thesis rohaida plus corrections

Durham E-Theses

The Relationship Between Governance Practices, Audit

Quality and Earnings Management: UK Evidence

BASIRUDDIN, ROHAIDA

How to cite:

BASIRUDDIN, ROHAIDA (2011) The Relationship Between Governance Practices, Audit Quality and

Earnings Management: UK Evidence, Durham theses, Durham University. Available at Durham E-ThesesOnline: http://etheses.dur.ac.uk/1382/

Use policy

The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission orcharge, for personal research or study, educational, or not-for-pro�t purposes provided that:

• a full bibliographic reference is made to the original source

• a link is made to the metadata record in Durham E-Theses

• the full-text is not changed in any way

The full-text must not be sold in any format or medium without the formal permission of the copyright holders.

Please consult the full Durham E-Theses policy for further details.

Academic Support O�ce, Durham University, University O�ce, Old Elvet, Durham DH1 3HPe-mail: e-theses.admin@dur.ac.uk Tel: +44 0191 334 6107

http://etheses.dur.ac.uk

THE RELATIONSHIP BETWEEN GOVERNANCE

PRACTICES, AUDIT QUALITY AND EARNINGS

MANAGEMENT: UK EVIDENCE

ROHAIDA BASIRUDDIN

A thesis submitted to Durham University in fulfilment of the

requirements for the degree of Doctor of Philosophy

DURHAM UNIVERSITY

BUSINESS SCHOOL

THE RELATIONSHIP BETWEEN GOVERNANCE PRACTICES, AUDIT

QUALITY AND EARNINGS MANAGEMENT: UK EVIDENCE

ABSTRACT

This thesis examines two empirical studies. Firstly, it examines the relationship

between corporate governance characteristics (relating to the size, composition of

independent members, financial expertise and meeting frequency of boards of

directors and audit committee) and audit quality. Secondly, the study investigates the

effectiveness of corporate governance characteristics and higher quality auditors in

constraining earnings management. There are three proxies of audit quality employed:

audit fees, non-audit fees and industry specialist auditors. Based on data obtained

from the FTSE 350 between 2005 and 2008, the first empirical findings suggest that

independent non-executive directors on board demand an additional and extensive

audit effort from the auditor in order to certify their monitoring function, resulting in

an increase in the audit fees and the perceived audit quality. The results also indicate a

positive relationship between independent board and non-audit fees, suggesting that

independent board support the view that the joint provision of audit and non-audit

services does not necessarily compromise auditor independence, but rather that it

broadens the auditors‟ knowledge and improves audit judgement. The findings from

the second empirical study suggest that higher quality auditors (which either charge

higher audit fees or are industry specialist auditors) are likely to reduce earnings

manipulation. However, no evidence suggests that NAS fees affect earnings

management. In addition, the current study finds inconsistent results linking the

corporate governance characteristics and opportunistic earnings. Overall, both

findings are consistent with agency theory, which states that independent board and

higher quality auditors are associated with effective monitoring, which in turn helps to

improve the quality of financial reporting. The findings are of potential interest to

policy makers, professionals and boards of directors, especially on issues relating to

audit quality and the mandating of corporate governance practices.

TABLE OF CONTENTS

ABSTRACT......................................................................................................... 2

TABLE OF CONTENTS................................................................................... 3

LIST OF FIGURES............................................................................................ 6

LIST OF TABLES.............................................................................................. 7

ABBREVIATIONS............................................................................................. 9

LIST OF APPENDICES.................................................................................... 12

DECLARATION................................................................................................. 13

ACKNOWLEDGEMENT.................................................................................. 14

CHAPTER 1: INTRODUCTION

1.1 Background of study................................................................................. 15

1.2 Motivation for the study........................................................................... 17

1.3 Contribution to existing knowledge.......................................................... 20

1.4 Structure of the thesis............................................................................... 25

CHAPTER 2:THEORETICAL FRAMEWORK

2.1 Introduction............................................................................................... 27

2.2 Definition of corporate governance.......................................................... 27

2.3 Agency theory and information asymmetries........................................... 29

2.3.1 The monitoring role of board of directors.................................... 31

2.3.2 The role of external audit............................................................. 33

2.4 Demand for external audit and different levels of audit quality............... 34

2.4.1 Monitoring hypothesis................................................................. 35

2.4.2 Information hypothesis................................................................. 36

2.4.3 Signalling or reputation hypothesis.............................................. 37

2.4.4 Insurance hypothesis.................................................................... 41

2.4 The association between board of directors, audit committee, external

auditor and financial reporting quality...................................................... 44

2.5 Summary................................................................................................... 50

CHAPTER 3: LITERATURE REVIEW AND HYPOTHESIS

DEVELOPMENT

3.1 Introduction.............................................................................................. 52

3.2 Definition of audit quality......................................................................... 52

3.3 A possible approach to measuring audit quality....................................... 58

3.3.1 Audit fees…………………......................................................... 60

3.3.2 NAS fees………………….......................................................... 63

3.3.3 Industry specialist auditors…………………............................... 66

3.4 The effectiveness of board of directors..................................................... 69

3.4.1 Board of directors size................................................................. 69

3.4.2 Independent board of directors..................................................... 70

3.4.3 Board of directors expertise......................................................... 71

3.4.4 Board of directors meetings......................................................... 72

3.5 The effectiveness of audit committee....................................................... 72

3.5.1 Audit committee size.................................................................... 73

3.5.2 Independent audit committee....................................................... 74

3.5.3 Audit committee expertise........................................................... 75

3.5.4 Audit committee meetings........................................................... 76

3.6 Earnings management............................................................................... 77

3.7 Accruals-based measure of earnings management................................... 80

3.8 The relationship between board of directors, audit committee and

demand for audit quality........................................................................... 82

auditor quality in constraining earnings management.............................. 94

3.10 Summary.................................................................................................. 104

CHAPTER 4: METHODOLOGY

4.1 Introduction............................................................................................... 113

4.2 Sample firms and period of study............................................................ 113

4.2.1 Sample selection for regression analysis...................................... 114

4.3 Sources of data.......................................................................................... 116

4.4 The definition and measurement of hypothesis variables........................ 117

4.4.1 The board of directors and audit committee variables................. 117

4.4.2 Audit quality variables................................................................. 117

4.4.3 Earnings management variable.................................................... 129

4.5 Model specifications and related control variables................................... 132

4.5.1 Audit fees model.......................................................................... 133

4.5.2 NAS fees model........................................................................... 136

4.5.3 Auditor industry specialist model................................................ 139

4.5.4 Earnings management model....................................................... 142

4.6 Data analysis procedures………………………………………………. 147

4.6.1 Descriptive statistics and correlation matrix............................... 147

4.6.2 Multivariate regression................................................................. 147

4.6.3 Further analysis and robustness tests............................................ 149

4.7 Summary................................................................................................... 150

CHAPTER 5: FINDINGS AND DISCUSSIONS - THE BOARD OF

DIRECTOR, AUDIT COMMITTEE AND AUDIT QUALITY

5.1 Introduction............................................................................................... 153

5.2 Descriptive statistics................................................................................. 153

5.3 Correlation matrix.................................................................................... 157

5.4 ANALYSIS I: AUDIT FEES

5.4.2 The additional analysis and robustness tests................................ 171

5.5 ANALYSIS II: NAS FEES

5.6 ANALYSIS III: INDUSTRY SPECIALIST AUDITORS

5.6.1 Univariate test.............................................................................. 202

5.7 Summary................................................................................................... 218

CHAPTER 6: FINDINGS AND DISCUSSIONS - THE

EFFECTIVENESS OF BOARD OF DIRECTORS, AUDIT

COMMITTEE, AUDIT QUALITY ON CONSTRAINING EARNINGS

MANAGEMENT

6.1 Introduction.............................................................................................. 222

6.2 Descriptive statistics and univariate test................................................... 222

6.3 Correlation matrix.................................................................................... 232

6.4 Multivariate regression............................................................................. 232

6.5 The additional analysis and robustness tests............................................. 242

6.7 Summary................................................................................................... 266

CHAPTER 7: SUMMARY AND CONCLUSION

7.1 Introduction............................................................................................... 268

7.2 Overview, summary and conclusion of the study.................................... 268

7.3 Implications of the study.......................................................................... 273

7.4 Limitations of the study............................................................................. 274

7.5 Recommendations for future research...................................................... 275

REFERENCES.................................................................................................... 277

APPENDIX.......................................................................................................... 299

LIST OF FIGURES

FIGURE 2.1 A typology of the theories relating to the roles of governing

board............................................................................................. 31

FIGURE 2.2 The investment in reputation and the premium for a higher

quality product.............................................................................. 40

FIGURE 3.1 A model of audit quality............................................................... 55

FIGURE 3.2 The key drivers of audit quality.................................................... 58

LIST OF TABLES

TABLE 3.1 Summary of the key literature relating to board of directors,

audit committee, audit quality and earnings management

studies........................................................................................... 106

TABLE 4.1 Sample selection procedures......................................................... 114

TABLE 4.2 Sample size and industry description for the first empirical

analysis – relationships between board of directors, audit

committee and audit quality......................................................... 115

TABLE 4.3 Sample size and industry description for the second empirical

analysis – relationship between board of directors, audit

committee, external auditor and earnings management............... 116

TABLE 4.4 The Big 4 auditor industry specialists (by year).......................... 125

TABLE 4.5 The Big 4 auditor industry specialist (by pooled)........................ 128

TABLE 4.6 Summary of all variables used...................................................... 150

TABLE 5.1 Descriptive statistics (N=674)....................................................... 160

TABLE 5.2 Pairwise correlation matrix (N=674) ........................................... 163

TABLE 5.3 The result of multivariate regression for audit fees model........... 170

TABLE 5.4 Heteroscedasticity test for audit fees model................................. 177

TABLE 5.5 VIF and tolerance values for audit fees model............................. 177

TABLE 5.6 The results of different estimators for audit fees model (N=674) 178

TABLE 5.7 The results of the multivariate regression of audit fees model

for “small” and “large” firms........................................................ 179

TABLE 5.8 Chow test.................................................................................... 179

TABLE 5.9 The results of audit fees model for the alternative test variable

definitions.................................................................................... 180

TABLE 5.10 The results of audit fees model with the additional control

variables........................................................................................ 181

TABLE 5.11 Endogeneity test for audit fees model.......................................... 182

TABLE 5.12 The results of 2SLS regression for audit fees model (N=674)..... 183

TABLE 5.13 The results of multivariate regression for NAS fees model........ 188

TABLE 5.14 Heteroscedasticity test for NAS fees model................................. 195

TABLE 5.15 VIF and tolerance values for NAS fees model............................. 195

TABLE 5.16 The results of different estimators for NAS fees model (N=674) 196

TABLE 5.17 The results of NAS fees model for “small” and “large” firms

(N=337)......................................................................................... 198

TABLE 5.18 The results of NAS fees model for the alternative test variable

definitions (N=674)...................................................................... 199

TABLE 5.19 The results of NAS fees model with the additional control

variables (N=674)......................................................................... 200

TABLE 5.20 Endogeneity test for NAS fees model.......................................... 201

TABLE 5.21 Univariate tests............................................................................ 204

TABLE 5.22 The results of multivariate regression for auditor industry

specialist model............................................................................ 206

TABLE 5.23 Heteroscedasticity test for auditor industry specialist model....... 211

TABLE 5.24 VIF and tolerance values for auditor industry specialist model... 211

TABLE 5.25 The results of different estimators for auditor industry specialist

model (N=674)............................................................................. 212

TABLE 5.26 The results of auditor industry specialist model for the

alternative test variable definitions (N=674)............................... 214

TABLE 5.27 The results of auditor industry specialist model with the

additional control variables (N=674)........................................... 215

TABLE 5.28 Endogeneity test for industry specialist model........................... 216

TABLE 5.29 The results of 2SLS regression for auditor industry specialist

model (N=674)............................................................................. 217

TABLE 5.30 The summary of the hypothesis and the findings – the

relationship between the corporate governance characteristics‟

and auditor quality........................................................................

TABLE 6.1 Descriptive statistics (N=613)....................................................... 225

TABLE 6.2 Univariate tests............................................................................. 227

TABLE 6.3 Pairwise correlation matrix (N=613)............................................ 229

TABLE 6.4 The results of multivariate regression for the earnings

management model...................................................................... 236

TABLE 6.5 Heteroscedasticity test for earnings management model.............. 246

TABLE 6.6 VIF and tolerance values for earnings management model.......... 247

TABLE 6.7 The results of GLS regression for the earnings management

model (N=613)............................................................................. 248

TABLE 6.8 Fixed or random effect estimator................................................. 254

TABLE 6.9 The results of fixed-effect/ random effect estimators for the

earnings management model (N=613)......................................... 255

TABLE 6.10 The results of earning management model for the alternative

test variable definitions (N=613).................................................. 261

TABLE 6.11 Endogeneity test for the earning management model.................. 261

TABLE 6.12 The results of 2SLS regression for earnings management model 264

TABLE 6.13 The summary of the hypothesis and the findings – the

relationship between the corporate governance characteristics‟

and auditor quality in constraining earnings management.......

ABBREVIATIONS

% Percentage

2SLS Two Stage Least Square

A&A Arthur Andersen

ACCA The Association of Chartered Certified Accountants

ACEXP Audit committee expertise

ACIND Audit committee independence

ACMEET Audit committee meeting

ACQ Number of acquisitions

ACSIZE Audit committee size

Adj. Adjusted

BDRNED Board of directors composition

BLOCK Institutional ownership

BRC The Blue Ribbon Committee

BRDEXP Board expertise

BRDMEET Board meeting

BRDSIZE Board size

CA Current asset

Cash Cash and cash equivalent

CEOs Chief Executive Officers

CFO Cash flow from operation scaled by lagged total asset

CL Current liabilities

DACCJM Discretionary accrual based on Jones model

DACCMJM Discretionary accruals based on Modified Jones

DACCROA Performance-adjusted discretionary accruals

DEBT Debt

DEP Depreciation and amortization expenses

DL Deloitte

e.g. Exempli gratia (Latin: for instance)

et al. et alia (Latin: and other)

EY Ernst & Young

FEERATIO1 Fee ratio of NAS fees to total fees

FEERATIO2 Fee ratio of NAS fees to audit fees

FORGN Foreign subsidiaries to total consolidated subsidiaries

FORGNSALE Proportion of the firm‟ foreign sales

FTSE Financial Times Stock Exchange

GAAP General Accepted Accounting Principles

GLS Generalize least square

GROWTH Growth rate in sales

H0 Hypothesis null

i Company i

i.e. id est (Latin: that is to say)

ICAEW The Institute of Chartered Accountants of England

and Wales

ICB Industry Classification Benchmark

INOWN Directors' ownership

KPMG KPMG Peat Marwick

L&H Laventhol and Horwath

LEVERG Proportion of debts to total assets

LIQ Ratio of current assets divided by current liabilities

LNAFEE Natural log of audit fees

LNASSET Natural logarithm of total assets

LNNAF Natural log of NAS fees;

LNTOTALFEES Natural log of the sum of audit and NAS fees

LOSS Negative income

MTBV Market to book value ratio

N Sample size

NAS Non-audit services

NEWDIR New appointment of external director

NWFUND Issuance of new shares or debt for cash

NWFUNDRATIO Ratio of new debt and equity issuance to total assets

OLS Ordinary least square

PWC Price Waterhouse Coopers

RECINV Proportion of total assets in account receivable and

inventory

RESTR Occurance of restructuring program

RETURN Fiscal year total stock return

ROA Return on assets

SEC Securities Exchange

SOX Sarbanes-Oxley Act of 2002

SPECLST_MS Industry specialist auditors - market share

SPECLST_MS30 Industry specialist auditors - cut off 30%

SPECLST_MSLEADER Industry specialist auditors - leader

SPECLST_PS Industry specialist auditors - portfolio share

SPECLST_WEIGHTED Industry specialist auditors - weighted average

SQSUBS Consolidated subsidiaries

t Period t or year-end of period t

TACC Total accruals

UK United Kingdom

US United States

VIF The Variance Inflation Factor

LIST OF APPENDICES

Appendix 1 The prohibitions of the specific non-audit services as outlined

in ES 5 (revised) – NAS provided to audited entities.................. 299

DECLARATION

I hereby declare that the materials contained in this thesis have not been previously

submitted for a degree in this or any other university. I further declare that this thesis

is solely base on my own research.

Rohaida Basiruddin

STATEMENT OF COPYRIGHT

“The copyright of this thesis rests with the author. No quotation from it should be

published without the prior written consent and information derived from it should be

acknowledged.”

Rohaida Basiruddin

ACKNOWLEDGEMENT

All praise be to Allah, the Lord of the Worlds, and may His Blessings and Peace be

upon our Prophet Muhammad. I thank Him for the blessings and for honouring me

with His guidance in order to complete this thesis.

I would like to express my sincerest gratitude to my supervisors, Dr. Aly Salama and

Professor Rob Dixon, for their guidance, knowledge, patience, time and support

throughout every stage of my PhD and the development of this thesis.

I thank Jenny Stewart (editor of International Journal of auditing), W. Robert Knechel

(editor of Auditing: A Journal of Practice & Theory) and four anonymous referees for

their helpful comments and suggestions on the early drafts of my papers. I also

appreciate the comments from the participants of the EAA 32nd

Annual Congress

Conference 2009 and BAA Doctoral Colloquium 2010. I extend my appreciation to

the academic and administrative staff of Durham Business School for their support

and assistance, especially Dr. Rebecca Stratling, Dr. Tareq Abd-Fattah, Dr. Amir

Michael, the doctoral office staff and the IT office staff.

I am also grateful to a number of PhD scholars: Nooraisah Katmun, Murya Habbash,

Prawat Benyasrisawat, Karuntarat Boonyawat, Amiruddin Muhamed and Siti

Uzairiah Md Tobi, for sharing their knowledge and time. I also want to thank all of

my Malaysian friends, especially Kak Wan, Kak Zu, Ustazah Hamidah and Azizah,

for their help and support to my family members during the period of my study.

Last, but not least, I am thankful to my parents, my parents-in-law and my siblings for

their continuous support and encouragement. A very special thanks to my husband,

Muhamad Rizal, and my three children, Qurratul‟ain, Qistina Ayuni and Qanitah

Atifa, for their unconditional love, endless support and patience. Thank you very

much for making my life so fulfilled.

CHAPTER 1

INTRODUCTION

1.1 Background of study

The issues of audit quality, corporate governance and earnings management have

received significant attention from government regulators, the auditing profession and

the public, especially following the recent high-profile corporate scandals. The

integrity of the financial reporting system is being questioned, the credibility of the

auditor is in doubt and a firm's governance system is liable to be blamed because of

the lack of auditor independence and oversight from the board. DeFond and Francis

(2005) claim that the consequence of the corporate scandal has renewed the

importance of independent audits and their linkage to the monitoring role of corporate

governance.

Agency theory provides an explanation of why independent audit is important to the

financial market. The independent audit helps to mitigate the agent-principal conflict

by providing the assurance that financial statements are carefully prepared and free

from material error (Wallace, 1980). It also reduces the likelihood of accounting fraud

and illegal reporting practices (Wallace, 1980), such as earnings management, so that

the market participant can use the financial reports without any hesitation. In addition,

the auditor can improve the quality of financial reporting through their competency

and willingness to report an accounting misstatement (DeAngelo, 1981) and to

respond to aggressive earnings conservatism1 (Ruddock et al., 2006). Furthermore, the

regulators believe that good corporate governance is able to improve the ability of

boards and their committees to manage effectively and in the best interest of

shareholders, whose trust and confidence is gained through higher quality auditing

(SOX, 2002; UK Corporate Governance Code, 2010).

1 Conservatism is defined as “the application of a higher standard of verification for favourable

information” (Ruddock et al., 2006: 706). According to Ruddock et al. (2006: 740) “conservatism is an

important attribute of financial reporting and is associated with the factors that underlie the demand for

audit quality”.

In most cases shareholders depend on the ability of a board and its committee to

monitor the independence of both the management and the auditors.2 Therefore,

responsibility for the quality of financial reporting is laid on the effective board and

its committee. Most prior studies have focused on the role of the audit committee as

the main agent in ensuring the integrity of financial information and dealing with

issues related to an external audit (Chen et al. 2005; Abbott and Parker 2000).

However, given that the board of directors is responsible for appointing and removing

the audit committee‟s members and the external auditors, their role is equally crucial

in promoting a higher quality of financial reporting. The Blue Ribbon Committee

(BRC, 1999: 6-7) reports that, “the performance of audit committees must be founded

in the practices and attitudes of the entire board of directors ... If the board is

dysfunctional, the audit committee likely will not be much better.” Similarly, a few

studies have indicated that the audit committee‟s effectiveness is associated with the

composition of the entire board of directors (Menon and William, 1994; Collier and

Gregory, 1999; Cohen et al., 2002; Boo and Sharma, 2008). Therefore, in this thesis,

while the demand for a higher quality auditor is recognised, the monitoring roles of

board and audit committee are argued to be the more important mechanisms by which

to promote a higher quality of financial reporting.

Specifically, two empirical studies will be examined in this thesis: (1) the study of the

relationships between the effective board of directors, the audit committee and audit

quality and (2) the study of the relationships between the effective boards, the audit

committee and auditor quality in respect of constraining earnings management. It has

been argued that the firms with effective boards and audit committees constantly

demand higher quality auditors because by employing higher quality auditors they add

credibility to the financial reports, increase the firm‟s value and are able to safeguard

themselves from damage to their reputations and legal exposure, all of which

promotes the shareholders' interests (Carcello et al., 2002). When market participants

lack the ability to directly observe the reported earnings, they may expect the

managers of the firms in a strong monitoring environment to engage less in earnings

manipulation. Therefore, the present study further argues that the firms with

monitoring mechanisms that consist of the board of directors and an audit committee

2 This referred to the minority shareholders.

with effective characteristics and a higher quality auditor have a greater ability to

constrain opportunistic earnings, hence reducing uncertainty in the reported earnings.

1.2 Motivation for the study

This thesis is motivated by three major considerations. Firstly, studies of audit quality,

earnings management and corporate governance continue to be important and to form

a part of regulators' and policy makers‟ concerns. Levitt (1998; 2000) claims that the

assessment of audit quality and earnings management is crucial because it is reflected

the investors‟ confidence in the financial reporting and it affects the allocation of

resources. A lack of investors‟ confidence in audit quality and reported earnings can

seriously undermine the financial market, since investors are the largest group of users

that provide capital support to the economic system. Despite the importance of audit

quality, the boards of directors and its committees are recognised as monitoring

mechanisms that may influence the quality of financial reports (Carcello et al., 2002;

Abbott et al., 2004; Larcker and Richardson, 2004; Lin and Hwang, 2009). Therefore,

how audit quality and the monitoring role of the board of directors and its committees

affect the market‟s perception of reported earnings and audit quality remains

important to the regulators and policy-makers.

Secondly, the non-audit services (hereafter NAS) continue to be controversial and are

viewed with scepticism due to the potential that they may compromise auditor

independence. In this thesis, NAS will be one of the proxies for audit quality. Prior

studies consistently suggest that the policy-makers, users and investors perceive the

NAS as impairing auditor independence (Panel on Audit Effectiveness, 2000; SEC,

2000; Beattie and Fearnley, 2002; Wines, 1994; Firth, 2002; Frankel et al. 2002;

Raghunandan, 2003; Sharma and Sidhu, 2001; Larcker and Richadson, 2004). For

example, in US, the Sarbanes-Oxley Act of 2002 (hereafter SOX) banned the auditor

from providing several services related to NAS.3 Consistent with SOX legislation, the

3 The Sarbanes-Oxley Act of 2002 was enacted on July 30th, 2002, applies to US public companies

(boards and management) and public accounting firms and has the objective of protecting investors by

improving the accuracy and reliability of corporate disclosures that are made pursuant to the securities

laws. Sarbanes-Oxley Act of 2002: Title II – Auditor Independence: Section 201: Services Outside

the Scope of Practice of Auditors. The public accounting firms are prohibited from performing the

following NAS for the financial statement audit clients: (1) bookkeeping or other services related to the

accounting records or financial statements of the audit client; (2) financial information systems design

and implementation; (3) appraisal or valuation services, fairness opinions, or contribution-in-kind

reports; (4) actuarial services; (5) internal audit outsourcing services; (6) management functions or

UK regulator also responds to NAS by issuing the Ethical Standard (ES) for auditors.

For instance, the ES 5 requires auditors to assess possible threats that may impair their

objectivity and independence, and to identify effective safeguards to reduce these

potential threats. In addition, ES 5 has also banned several NAS services that were

believed to have compromised auditors‟ objectivity and independence.4 In summary,

both regulatory agents claim that the higher provisions of NAS can impair auditor

independence. Despite the negative effect of NAS on auditor independence, several

studies argue that the joint provision of audit and NAS may also broaden the auditor‟s

knowledge and improve the audit judgement. This, in turn, enhances the quality of

financial reporting (see Simunic, 1984; Beck et al., 1988a; Arruñada, 1999a; 1999b;

2000; Wallman, 1996; Goldman and Barlev, 1974). Such mixed arguments motivate

the present study to examine the levels of NAS fees and their relation to corporate

governance and earnings management.

Furthermore, DeFond and Francis (2005) call for research on NAS studies to

incorporate fees dependency as an alternative measure of auditor independence. They

claim that, during the drafting of the SOX legislation, the study carried out by Frankel

et al. (2002) has been used to generalize the implication of higher levels of NAS fees,

and that subsequent studies in NAS fees have claimed that Frankel's research was

sensitive to the research design, sample selection and model specification.5 They

further argue that the SEC had never raised the issues of fee dependency when they

proposed to ban NAS in 2000. Hence, DeFond and Francis (2005) call for research of

NAS studies to take into account the total fees (the sum of audit and NAS fees) as an

alternative measure of the financial dependencies that are believed to affect the

auditor's objectivity. As far as the present study is concerned, no studies in the UK

have employed the total amount of auditor fees in examining the relationships

between corporate governance and NAS, and between the corporate governance and

NAS in constraining earnings management. Therefore, the total fees will be one of the

NAS measures to fill the gap in the UK literature. In fact, there are three other

human resources; (7) broker or dealer, investment adviser, or investment banking services; (8) legal

services and expert services unrelated to the audit; and (9) any other service that the Board determines,

by regulation, is impermissible. 4 The prohibitions are set out in Appendix 1.

5 Frankel et al. (2002) suggest that the firms that paid higher NAS fees are likely to have higher

abnormal accruals, indicating the managerial discretion in managing earnings.

measures that will be employed in this thesis (i.e. the level of NAS fees compared to

the sum of audit and NAS fees, the ratio of NAS fees to total fees and the ratio of

NAS fees to audit fees) in order to ensure that the results are robust to several

measures of auditor independence.

Thirdly, most of the prior studies relating to audit quality, corporate governance and

earnings management have been conducted in the US. This offers different

institutional settings from the UK market and thus limits the generality of their

findings for contexts beyond the US. Although the UK and the US share some

common features, there are differences in corporate governance systems (e.g. Toms

and Wright, 2005). In the UK, share ownership is less dispersed than in the US and

the nature of investor activism differs and is shaped by the different rights of

shareholders in the two countries (Kirchmaier et al., 2005). As noted by Aguilera et

al. (2006), British institutional investors (mainly pension funds and insurance

companies) tend to be more dominant than their American counterparts (mainly

mutual funds). Also, major differences exist regarding the preparation of financial

statements and the US disclosure system is more demanding. The US firms are

required to disclose more detailed information about audit committees and auditors

than are the UK firms (Lennox, 2003). Another area of divergence is the constraints

on the exercise of board leadership (Aguilera et al. 2006). Most of the UK listed firms

separate the role of the Chairman and the Chief Executive Officer (CEO), while most

of the American CEOs are also the Chairman of the Board (Higgs Report, 2003).

Also, CEOs‟ pay and stock-based incentives in the US are relatively higher than in the

UK (Conyon and Murpht, 2000).

The UK adopts a voluntary corporate governance system that operates on a „comply

or explain‟ basis. The system is based on a set of good corporate governance practices

to which listed firms are expected to adhere or, where they do not, they are expected

to provide an explanation for their non-compliance in their annual report.6 This is

different from the mandatory corporate governance regime adopted in the US under

6 The UK Corporate Governance Code (formerly the Combined Code) was issued in 1998 and has been

revised since then. The fundamental changes in the Code were made in 2003 based on the

recommendations set out in the Higgs Report (which reviewed the role and effectiveness of non-

executive directors) and the Smith Report (which provided guidance on audit committees‟ functioning).

SOX. Indeed, the voluntary corporate governance system in the UK is one of the most

comprehensive and it offers more specific guidelines on the formation of a board and

its committee when compared with other countries that adopt a similar approach

(Sharma et al., 2009). The voluntary formation allows the firms to choose the optimal

structure of governance system to fit their needs. This could be different in the US

where the firms are forced to follow the SOX requirements without any consideration

of how they might prioritise their needs and raise concerns over the cost-benefit of

SOX on smaller firms.

Furthermore, the litigation environment in the UK differs from that in the US market

and other regimes and, therefore, it has a different effect on an auditor‟s reputation

and performance (Khurana and Raman, 2004). When auditors‟ reputations are less

likely to be affected by the regulator or by litigation, there is also less incentive for

them to perform high-quality audits (Francis, 2006; Khurana and Raman, 2004).7

Therefore, in the UK, the current study expects to find it highly desirable that, in their

monitoring roles, the board and its committee ensure the quality of an auditor‟ work is

not jeopardised by the lower litigation environment.

In summary, given that the US market offers different institutional settings,

governance structures and litigation environments from those in the UK, the

generalizability of American findings is limited. For these reasons, the present study

examines the association between governance practices, audit quality and earnings

management in the UK.

1.3 Contributions to knowledge

This thesis represents a comprehensive study on audit quality, corporate governance

and earnings management, particularly in the UK market. Using the current data of

FTSE 350 firms for the fiscal years of 2005 to 2008, the first part of this thesis

examines the relationship between corporate governance and audit quality. The

second part provides evidence linking audit quality, corporate governance and

7 Khurana and Raman examine the relationship between cost of capital, litigation risk and financial

reporting quality in the US and conclude that firms employing the big-4 auditors quantitatively have

lower cost of capital than those with non-big 4 auditors but not in Australia, Canada and the UK due to

lower litigation environment in these countries. They conclude that rather than brand name, the

litigation risk drives the audit quality differences.

earnings management. Audit quality is measured using several proxies such as audit

fees, NAS fees and industry specialist auditors. Corporate governance mechanisms

are associated with the effective characteristics of the board and audit committee and

they include committee size, composition of independent members, financial expertise

and frequency of committee meetings.

Several contributions to knowledge are made through this thesis. Firstly, it contributes

to debates on the importance of audit quality and corporate governance issues

subsequent to the recent audit failure scandals. The findings from the first empirical

investigation suggest that independent non-executive directors on board demand an

additional and extensive audit effort to certify their monitoring function, resulting in

an increase in the audit fees and the perceived audit quality. The second empirical

finding suggests that the higher auditor fees and industry specialise auditor are

associated with reduced earnings manipulations. Together, both documented findings

support the proposition of agency theory and the regulatory concern that an effective

board of directors and higher quality auditors are associated with effective monitoring

and improved quality of financial reporting.

Secondly, this thesis contributes to the growing literature on studies of audit quality,

corporate governance and earnings management. As stated in the motivations for the

study, prior research in these areas has predominantly been undertaken in the US,

which offers a different litigation environment, institutional setting, governance

structure and auditor incentive, thus limits the generalizability of the findings to other

countries. In particular, the current study expands the prior literature in six areas:

(1) To the best of the author's knowledge, no studies have examined the

relationship between the industry specialist auditor as proxy for audit quality

and the effective characteristics of the board and audit committee in the UK.

Previous studies in this area were carried out by Abbott and Parker (2000),

Beasley and Petroni (2001) and Chen et al. (2005) using US and Australian

samples. The investigation of UK firms expands the existing literature by

providing evidence from voluntary governance practice and different

institutional settings and litigation environments, each of which, it is argued,

drive quality differences in the audit market.

(2) None of the prior studies on how the auditor industry specialist affects

corporate governance and earnings management has used the complementary

approach to calculate the effect of the auditor industry specialist. Most of these

studies utilized the market share and portfolio approach (Abbott and Parker,

2000; Beasley and Petroni, 2001; Chen et al., 2005; Krishnan, 2003a; Balsam

et al., 2003). According to Neal and Riley (2004) a complementary approach

captures the complementary effect of both the market share and the portfolio

approach and offers a solution for inconsistencies between these two main

approaches.

(3) Previous UK studies that examine the relationship between corporate

governance and auditor fees (audit fees and NAS fees) have been undertaken

by Collier and Gregory (1996), O‟Sullivan (1999; 2000), O‟Sullivan and

Diacon (2002), Adelopo (2010) and Zaman et al. (2011). Specifically, Collier

and Gregory investigate the effect of the establishment of an audit committee

on audit fees, using the 1991 data, while O‟Sullivan examines the

establishment of an audit committee and the effect of the composition of

independent members on board and audit committee on audit fees, using data

from 1992 and 1995. Recently, Adelopo (2010) examines more

comprehensive corporate governance characteristics using single and

simultaneous equations of audit and NAS fees on data from 2005 and 2006.

Zaman et al. (20011) investigate the influence of audit committee

effectiveness and several board characteristics (independent board, meetings

frequency and duality role) on audit fees and non-audit fees using data 2001 to

2004. However, all of these studies do not consider financial expertise of the

board of director. Previous US studies suggest that the boards of directors that

are financially literate improve the quality of financial reporting (Xie et al.,

2003; Agrawal and Chadha, 2005; Lee, 2008). By examining the effect of the

financial expertise of the board of directors on audit fees and NAS fees, this

thesis extends prior research on the effect of financial literacy of board

members on the auditor remuneration in the UK market.

(4) In relation to the studies examining the effect of corporate governance and

auditor quality on constraining earnings management in the UK, prior studies

have been undertaken by Ferguson et al. (2004), Peasnell et al. (2000; 2005),

Antle et al. (2006), Kwon et al. (2007), Habbash et al. (2010), Sun et al.

(2010) and Habbash (2010). Ferguson et al. investigate the NAS fees, the big-

5 auditors and several board characteristics (e.g. non-executive directors on

board and CEO duality roles) on earnings management using the data from

1996 to 1998, while Peasnell et al. examine the board of director

characteristics (e.g. board size, non-executive directors on board, CEO

duality), the establishment of audit committee and the big-5 auditors using the

data in the period between 1991 and 1996. Antle et al. (2006) look at the joint

determination of audit fees, NAS fees and discretionary accruals using data

from 1994 to 2000, while Kwon et al. (2007) investigate how a country‟s legal

system affects the industry specialist auditor in constraining earnings

management in 28 countries including the UK. Habbash et al. (2010)

investigate the commitment of independent directors (i.e. composition,

meetings and directors fees) and number of board meeting , while Sun et al.

(2010) only control the size of boards and audit committee meetings in their

earnings management model. All of these studies exclude the financial

expertise of board and most of audit committee characteristics (e.g. size,

composition, financial expertise and meeting frequency) in examining the

effect of corporate governance on earnings management. Recently, Habbash

(2010) examines the corporate governance characteristics (board and audit

committee characteristics) and auditor quality variables separately using two

different earnings management models. In his first model, he examines only

the board of director characteristics, while in second model he investigates the

audit committee attributes and auditor variables. The current study fills the gap

between all these studies by examining the characteristics of board, audit

committee and auditor quality variables on the single earnings management

model. Larcker and Richardson (2004) claim that when a firm's corporate

governance and auditor quality are isolated from one and another, it may result

in an incomplete analysis of earnings management because the monitoring role

of auditors varies depending on the strength of the firm‟s corporate

governance.

(5) In this thesis, audit quality is measured using several proxies: audit fees, NAS

fees and industry specialist auditors. Even each proxy is measured using

several approaches. For example, the NAS fees proxy consists of the natural

log of NAS fees, the natural log of total fees (sum of audit and NAS fees), the

ratio of NAS fees to total fees and the ratio of NAS fees to audit fees. While

the industry specialist auditor is measured in five ways, three measures are

continuous variables (equal to the respective auditor's market share and the

auditor‟s portfolio share, and complimentary between the market share and the

portfolio share) and two measures are dichotomous variables (the industry

leader and the auditor‟s market share at a 30 percent cut-off point in each

particular industry). By examining multiple proxies of audit quality and

several measures for each proxy, the current thesis provides an analysis of the

impact of corporate governance characteristics and earnings management that

is more comprehensive than prior studies that mostly examined only a single

proxy of audit quality.

(6) The present study contributes to the methodology literature by considering the

analysis using several estimations including the OLS regression, the robust

regression, the least square regression with robust standard error, the quantile

regression, the probit regression and the heteroskedastic ordinal regression.

Several estimators are employed to ensure the efficiency of the data analysis

since the OLS regression may not be an efficient estimator when some of the

assumptions are not fulfilled. In addition, none of the prior research in

corporate governance and audit quality employs the heteroskedastic ordinal

regression that argues to increase the efficiency of logit/ probit regression in

the presence of heterscedasticity. Further, all the analyses take into account

endogeneity issues that have been neglected in a few previous studies.

Lastly, this thesis contributes to the debate on the joint provision of audit and NAS.

The result from the NAS model suggests that the firms with a higher provision of

independent directors on the board are likely to be associated with higher NAS fees.

This suggests that an independent board may support the higher provision of NAS that

is likely to improve the quality of audits due to the presence of knowledge spillover

effects. Furthermore, no evidence suggests that the NAS is associated with

opportunistic earnings. This finding contradicts the regulatory concern that the

provisions of NAS compromise auditor independence and, thereby, reduce the quality

of financial reporting.

1.4 Structure of research

This thesis is structured into seven chapters. This chapter discusses the background of

the study, the motivation and the contribution that it makes.

Chapter 2 focuses on the theoretical framework underpinning this study. The main

discussion centres on agency theory as the primary theory, which suggests that the

monitoring roles of the board of directors and the audit committee, as well as the

demand for independent audits, help to mitigate the agency conflict. The importance

of independent audit and different levels of audit quality for the firms and market

participants is explained through several hypotheses such as: the monitoring

hypothesis, the information hypothesis, the signalling hypothesis and the insurance

hypothesis. The relevant explanations as to why the board of directors and its

committee and the auditor demand different levels of audit quality and limit

opportunistic earnings are also highlighted.

Chapter 3 reviews the prior studies related to the three groups of subjects: audit

quality, corporate governance and earnings management. Under the audit quality

reviews, the current study defines audit quality and its possible measures. The

effective characteristics of a board of directors and audit committee, including the size

of the committee, the composition of independent members, the financial expertise of

members and the frequency of meetings are also reviewed and discussed based on the

propositions of agency theory and on prior documented evidence. The reviews on

earnings management take into account the definition of earnings management, the

motivation of opportunistic earnings and the measurement of earnings. Previous

studies that are related to the association between audit quality and corporate

governance and between the effects of corporate governance and auditor quality on

constraining earnings management are also discussed. The end of the chapter follows

the development of the hypotheses by previous authors.

Chapter 4 explains the methodology employed in this study. It explains and justifies

the sample firm selection, the period of the study, the definitions and the

measurements of the main variables, hypothesis variables, model specification and

related control variables. The description of the source of data, the data collection

procedures and analysis procedures are also discussed.

Chapter 5 and 6 present the results of the empirical findings on the associations

between corporate governance and audit quality, and the relationship between

corporate governance, audit quality and earnings management, respectively.

Finally, chapter 7 provides an overall summary and concludes the study. The

implications, limitations and the recommendations for future research are discussed in

this final chapter.

CHAPTER 2

THEORETICAL FRAMEWORK

2.1 Introduction

This chapter provides the theoretical framework for the present study. The main

discussion is on the agency theory that highlights the relationship between principals

and agents and the conflict that arises between them because of the different goals. As

part to reduce the agency conflict, the monitoring role of board of directors and audit

committee as well as the external audit are demanded. Several hypotheses related to

the demand for audit and different levels of audit quality are also discussed in this

chapter. The association between board of directors, audit committees, external

auditors and financial reporting are highlighted. Finally, the summary and conclusion

are presented in last section.

2.2 Definition of corporate governance

So far, existing studies have indicated that there is no specific acceptable definition

for corporate governance (e.g. Solomon, 2007). However, several definitions of

corporate governance have been discussed in prior studies (e.g. the Cadbury Report,

1992; Donaldson and Preston, 1995; Shleifer and Vishny, 1997; Turnbull, 1997). For

example, the Cadbury Report (1992: 15) defines corporate governance as “a system

by which companies are directed and controlled.” This definition highlights the roles

of the main players in an organisation that is comprised of shareholders, a board of

directors and the auditor. As cited in the Cadbury Report (1992), the shareholders are

responsible for appointing directors and auditors and for ensuring that the appropriate

governance system is in place. The directors‟ function is associated with how the firm

is governed, while the auditors‟ main role is to provide an independent check on

financial statements to shareholders.

Shleifer and Vishny (1997: 737) describe corporate governance as “dealing with the

ways in which suppliers of finance to corporations assure themselves of getting return

on their investment”. They suggest that legal protection of investor rights and

concentrated ownership helps to control management discretion so that financiers are

able to get returns on their investments.

Consistent with the Cadbury Report (1992), that stresses how firms are directed and

controlled, Denis (2001: 192) states that corporate governance “encompasses the set

of institutional and market mechanisms that induce self-interested managers (the

controllers) to maximize the value of the residual cash flows of the firm on behalf of

its shareholders (the owners)”. Alternatively, Solomon (2007: 14) considers

stakeholder concerns in the definition of corporate governance which is perceived as a

“system of checks and balances, both internal and external to companies, which

ensures that companies discharge their accountability to all their stakeholders and act

in a socially responsible way in all areas of their business activity.”

These various definitions and explanations of corporate governance exist because the

authors view corporate governance from different perspectives and through different

theoretical frameworks. For example, the definitions of corporate governance that are

outlined by Cadbury (1992), Shleifer and Vishny (1997) and Denis (2001) seem to

agree that corporate governance is associated with ownership and control and that it is

aimed at maximising the shareholders‟ wealth. These definitions are influenced by

agency theory. On the other hand, Solomon‟s definition is consistent with stakeholder

theory which believes that in addition to maximising the shareholders‟ wealth, social

and environmental issues are of significant importance to the firm. Stakeholder theory

acknowledges that individuals, both inside and outside a firm, such as employees,

suppliers, customers, publics, governments or other individuals or groups may have

an effect on or be affected by the actions of that firm. These groups of individuals are

referred as stakeholders (Freeman, 1984). The firms are responsible to carry out the

actions that not only benefit them, but that benefit society as a whole.

Given that the current study examines the effect of the roles of the board of directors,

the audit committee and the external audit on financial reporting quality (i.e. earnings

management and audit quality), for the purposes of the study, corporate governance is

viewed as a monitoring system of checks and balances to ensure that the interests of

shareholders are safeguarded. Such a view can be expressed appropriately in agency

theory. In addition, the focus of the current study is related to the quality of financial

statements rather than on a firm's impact on social and environmental factors,

therefore, stakeholder theory seems to deviate from the aim of the study.

2.3 Agency theory and information asymmetries

The proposition of agency theory has been addressed by Jensen and Meckling (1976)

by the introduction of the concept of agency cost. They apply the concept of agency

cost to explain issues associated with the separation of ownership and control in a

large corporation, consistent with Berle and Means‟ (1932) propositions.

The ultimate element in agency theory is the conflict of interest between principals

and agents. A principal (shareholder) assigns the power of the decision maker to an

agent (manager) who, as an agent, executes their duties on behalf of the principal

(Jensen and Meckling, 1976). Conflicts and dissimilar interests lead to information

asymmetries between the two parties. The existence of information asymmetries

results in two major agency problems, namely, moral hazard and adverse selection

problems.

Moral hazard problems are associated with the problem of hidden actions when agents

have the incentive to pursue self-interested behaviour. They arise when principals are

unable to observe actions that are undertaken by the agents. Formally, an agent is

expected to maximize the principal‟s wealth through their actions and decisions.

However, agents tend to pursue their own interests. By contrast, adverse selection

problems are associated with hidden information, where the agent has more

information than the principal. Both problems may create, for example, the

phenomenon of earnings management that, in turn, may cause shareholders, investors

and debt holders to be unable to distinguish the true economic value of a firm.8

According to agency theory, since the managers or agents are inspired by extrinsic

motivations (Sundaramurthy and Lewis, 2003), the principals have to identify ways to

motivate the agents and to ensure that they act in the best interest of the principals.

Jensen and Meckling (1976) suggest that agency cost can be an alternative way to

reduce agency conflict and they define agency cost as consisting of monitoring cost,

bonding cost and residual loss. Monitoring costs are the costs that are associated with

the appointment of appropriate agents, such as external auditors, and with

mechanisms that control the agents‟ behaviour, such as the roles played by the board 8 Earning management is defined as “in the sense of purposeful intervention in the external financial

reporting process, with the intent of obtaining some private gain” (Schipper, 1989: 92).

of directors. Bonding cost is the cost that is associated with contracting in order to

ensure that agents always make decisions that support the principal‟s wealth. These

costs include those that are related to the agent‟s compensation system. Residual loss

is the agency loss that is associated with the imbalance between monitoring and

bonding costs or, in other words, it is the reduction in principals' welfare that arises

from an imperfect alignment of interest between agents and principals (Jensen and

Meckling, 1976).

In this thesis, the monitoring roles of board and auditor are studied as mechanisms

that mitigate agency conflicts. The board of directors acts on behalf of the

shareholders and represents the shareholders‟ interests through overseeing managerial

functions. Zahra and Pearce (1989) suggest that agency theory is the most

comprehensive theory that clarifies the board of directors‟ functions and that

highlights the importance of their controlling role. Consistent with this notion, Hung

(1998) also claims that agency theory is a convincing theory for explaining the

boards‟ monitoring role.

Alongside the monitoring role of a board of directors, Solomon (2007: 137) claims

that the external audit represents another crucial element of a firm‟s internal control

system and that it provides a check and balance system that helps shareholders to

monitor and control the managements‟ activities. As pointed out by the Cadbury

Report (1992: 36),

“The annual audit is one of the cornerstones of corporate governance.

Given the separation of ownership from management, the directors are

required to report on their stewardship by means of the annual report and

financial statements sent to the shareholders. The audit provides an

external and objective check on the way in which the financial statements

have been prepared and presented, and it is an essential part of the checks

and balances required.”

Therefore, it can be argued that agency theory is essential to the present study since it

recognizes the monitoring roles of board of director and an external audit as

mechanisms to control management behaviours. The following sections explain the

characteristics of board and audit committee that contribute to the effectiveness of

their monitoring function and the role of an external audit.

Figure 2.1: A typology of the theories relating to the roles of governing boards

(Source: Hung, 1998)

2.2.1 The monitoring role of board of directors and audit committee

Fama and Jensen (1983) suggest that boards of directors are the primary decision

makers in an organisation and that they have the power to compensate entire decisions

that are made by the top management. Fama and Jensen also propose that, in the

decision making process, the initiation and implementation should be separated from

the ratification and monitoring of decisions in order to ensure that monitoring

functions are more efficient.9

9 The “initiation and implementation” of decisions are known as decision management, while

“ratification and monitoring” are known as decision control (Jensen and Meckling, 1983: 304).

In other words, agency theory suggests that, in order to ensure that effective

monitoring functions are in place, board of director members should include a

representative from outsider members (hereafter non-executive directors) who is

independent from management.10

Vance (1983: 46) adds that the independent non-

executive directors provide unbiased assessment that is “stockholder-oriented” and

that establishes a best practice of “check and balance” on management's actions. The

non-executive directors are also important because they possess significant knowledge

(e.g. “capital market, technology, corporate law”) that enables them to complement

insider information and act as arbitrators in any conflict that may arise between the

insiders (Fama and Jensen, 1983: 314). In summary, the independent non-executive

directors are better at monitoring management, due to their „independent‟ and

„complimentary knowledge‟ characteristics.

The proposition of independent non-executive directors in agency theory is

contradictory to the principal of stewardship theory. Stewardship theory suggests that

the manager is acting as a steward and that their endeavours promote their principals‟

interest (Donaldson and Davis, 1991; Davis et al., 1997). The managers are inspired

by non-financial motivations and intrinsic gratification resulting from hard work and a

challenging working environment (Donaldson and Davis, 1991). As stated by

Donaldson and Davis (1991: 51), the manager, “far from being an opportunistic

shirker essentially wants to do a good job, to be a good steward of the corporate

assets”. In order to maximize the potential of the managers, the appropriate approach

is to establish an empowering structure (Donaldson and Davis, 1991). Managers are

supposed to be given clear instructions and a higher position in the organisation's

hierarchy where they would have autonomy and authority to make decisions and they

would be enabled to use their full capacity in achieving the organisation‟s objectives.

From a shareholder's point of view, the dominant insiders (hereafter known as

executive directors) on boards are preferred to non-executive directors. This is

because they have greater knowledge and awareness of current operations and they

Specifically, the outsider directors, however, can be distinguished between those who are

independent from management and have no relationship that would effect the exercising of their

judgments and decision making in the organization and those who are not (Lawrence and Stapledon,

1999). The non-executive director who is related to management is also been known as gray or

affiliated non-executive director. The independent non-executive director is believed to have better

monitoring position than gray or affiliated non-executive director (Lawrence and Stapledon, 1999)

have more technical expertise and, hence, they assume a more responsible attitude

towards the organisation (Muth and Donaldson, 1998). Therefore, the shareholders

could expect more return from them than from non-executive directors who are

assumed to be less informed about the organisation and to have a self-serving attitude.

Although stewardship theory identifies that executive directors are more beneficial

than non-executive directors, the present study believes that the monitoring role of a

board of directors as described by agency theory is more relevant to explaining

variations in audit quality and earnings management. Agency theory recognises the

independence of non-executive directors as a monitoring mechanism which is vital to

the promotion of high quality audits and higher quality financial reporting. Hung

(1998) suggests that the executive‟s task is focus on a „strategic role‟ rather than a

„monitoring role‟. In real systems, the agents or stewards are intent on pursuing their

own interests instead of the interests of others.

As well as having an independent non-executive director on a board and in an audit

committee, empirical evidence also suggests that the size of a committee, specific

knowledge, experience and a greater frequency of meetings may strengthen a board

and audit committee‟s monitoring functions (e.g. Lipton and Lorsch, 1992; Yermark

1996; Chen and Zhou 2007; Monks and Minow 2008; Dezoort 1998; Carcello et al.

2002; Abbott et al. 2003a, Krishnan and Lee 2009; Menon and Williams 1994; Vafeas

1999; Abbott et al. 2004; Ronen and Yaari 2008). Zahra and Pearce (1989: 308-310)

claim that the efficiency of a board of director‟s function is dependent on the (1) size

of board and the type of membership, (2) the attributes of the directors, such as their

competence and skills, (3) the establishment of the appropriate committees and (4) the

board of director‟s meeting (e.g. communication between directors, agenda and

documentation). Walker (2004: 158) also notes that “the performance of audit

committees necessarily depends on the people involved, their knowledge, skills,

critical capacities, skepticism and determination”. The empirical evidence for each of

these characteristics will be discussed later in Chapter 3.

2.2.2 The role of external audit

Audit functions have been observed to exist since the 13th

century to provide a form

of assurance that the financial information provided by the management accurately

represents the financial position of the firms (Watt and Zimmerman, 1983).11

Consistent with agency theory, audit function is viewed as a mechanism to reduce

uncertainty on the levels of information asymmetry between shareholders/investors

and management. As shareholders and investors have limited access to internal

information from within a firm, the independent audit reports on the truth and fairness

of financial statements that are prepared by management. As outlined in the

International Standard of Auditing, ISA (UK and Ireland) 200: Overall Objective of

the Independent Auditor and the Conduct of an Audit in Accordance International

Standards on Auditing (UK and Ireland), (ABP, 2009: 2-3):

“The purpose of an audit is to enhance the degree of confidence of

intended users in the financial statements. This is achieved by the

expression of an opinion by the auditor on whether the financial

statements are prepared, in all material respects, in accordance with an

applicable financial reporting framework… As the basis for the auditor‟s

opinion, ISAs (UK and Ireland) require the auditor to obtain reasonable

assurance about whether the financial statements as a whole are free from

material misstatement, whether due to fraud or error. Reasonable

assurance is a high level of assurance.”

Variations in the level of conflict and information asymmetry are assumed to differ

from firm to firm and may demand different levels of auditing and of audit quality

(DeAngelo, 1981; Watt and Zimmerman, 1986). The higher the agency cost, the

larger the information asymmetries‟ gap and thus the higher the levels of audit quality

will be demanded. The next section explains the relevant hypotheses in order to

clarify why a firm's management, including the board of directors, audit committee,

shareholders and investors, demands audit services and different levels of audit

quality.

2.3 Demand for external audit and different levels of audit quality

Several hypotheses have been used in the prior literature to explain the demand for

audit services and the different levels of audit quality. Each of these hypotheses is

now reviewed. It is important to highlight that these hypotheses may appear to be

The audit function exists as early in the development of business corporations since 1220,

accomplished by the directors or shareholders of the companies. In between early and mid nineteenth

century in UK and US, the audit then started been prepared by the professional auditors due to the

increasing numbers of corporations and complexity of the accounts which required the skill and the

expertise of professional auditors (Watts and Zimmerman, 1983).

related to one and another (e.g. Wallace, 1984; Willenborg, 1999; Menon and

William; 1994).

2.3.1 Monitoring hypothesis

The monitoring hypothesis is based on the agency relationship. Agency theory

suggests that agency cost is a potential solution to agent-principal conflicts and that

one of the answers to the problem is provided by an independent audit (Jensen and

Meckling, 1976). According to Wilson (1983), the monitoring role of audit minimizes

the moral hazard and adverse selection problems that arise from the information

asymmetries problem.12

As cited in Wilson (1983), in the case of a moral hazard

problem, the managers responsible for protecting a firm's assets may misuse the assets

or fail to maintain them, in which case such actions are not directly observable by the

owner and potential investors. In the case of an adverse selection problem, such assets

have their own fixed values. The managers have more information about these values

and they are able to manipulate information for their own personal gain.

Consequently, owners need to adopt an effective way to monitor managers‟

opportunistic behaviour and the credibility of the information provided by managers

as well as considering how to improve the investors‟ opportunities to observe such

assets.13

One possibility of achieving this is through independent audits. Auditors

provide managers and potential investors with reliable verification and information on

the value of assets. In other words, the independent audit provides an assurance to the

owners and potential investors that the information provided by managers is reliable.

The independent audit can be generated by the principal (monitoring cost) or the agent

(bonding cost). Humphrey (1997) argues that agents can demand an independent audit

because principals normally tend to neglect monitoring activity as they are able to

safeguard themselves from the risk of loss by paying lower wages to an agent (subject

to the cost of an independent audit being less than the wage loss that an agent could

suffer without an independent audit). The presumption is that principals will pay more

The monitoring role also can be regarded as assurance role since Brown et al. (2008) suggest that the

primary purpose of audit is to reduce the information asymmetries between managers and stakeholders

which, consistent to Wilson (1983) that suggest the independent audit able to mitigate the moral hazard

and adverse selection problems.

13 The owners are able to improve the accuracy and the reliability of the information to the potential

investors via the appointment of reputation auditors (Wilson, 1983).

to agents for work that has been verified by an independent audit than that which has

not been so verified. As pointed out by Wallace (1980: 14), “the stewardship

(monitoring) hypothesis states than when one party is delegated decision making

power, he has an incentive to agree to be checked if the benefits from such monitoring

activities exceed the related cost.”

According to Wallace (1980) an independent audit provides an assurance that the

financial reports that are provided by the managements have been carefully prepared

and they are free from material errors. Thus, the market participants including the

potential investors can use the audited financial statements without any hesitation.

Moreover, an independent audit also mitigates financial statement fraud and illegal

reporting and improves the internal control and operational efficiency of a firm

(Wallace, 1980; Chow 1982). For example, when managers know that their financial

reports will be examined by auditors, fraud and illegal behaviour can be minimised

indirectly because they are worried that such actions will be discovered by the

auditors. In addition, when auditors perform an audit review or audit testing on the

internal control system of a firm, they will discover if certain internal control

procedures are missing or have not been performed properly. Thus, auditors typically

provide recommendations to improve existing internal control systems. Such

restrictions and recommendations are able to improve the effectiveness and efficiency

of a firm‟s operation. In summary, these observations suggest that audit services not

only provide a monitoring tool for owners, managers and potential investors, but also

for the whole organization including its employees and creditors.

2.3.2 Information hypothesis

As previously stated, the higher the agency conflicts, the larger the information

asymmetries and thus the higher the quality of audit services that will be demanded.

Wallace (1980) suggests that investors demand audited financial statements because

the quality of financial information is improved through an independent audit. He

further suggests that audited financial information is able to (1) reduce market-related

(systematic) and firm-specific (unsystematic) risks, (2) improve decision making and

(3) provide access to new information for investors. As cited in Wallace (1980: 16-

17), a risk-averse investor may demand a higher rate of return for the higher levels of

risk or pay a higher risk premium to reduce the levels of uncertainty or investment

risk. It is assumed that the risk premium is associated with an individual investor's

assessment of an audit service; through audit, uncertainty about the accuracy of

financial information provided by management can be reduced (Shakun, 1978). If the

sum of risk premium for each investor is mutually adjoined exceeds the cost of audit,

the audited financial information is beneficial to all parties since all parties enjoy less

uncertain information. According to Wallace (1980), some investors may also reduce

their investment risk by developing a portfolio of both audited and unaudited

investment opportunities. Any reduction in the risk premium that is linked to the

audited information will be compensated through a specific firm‟s audit cost.

However, the unaudited investment portfolio may cause an increase in the variability

of the market and thus the cost of audits could be balanced against the demand on the

unverifiable market risk premium. Moreover, the barriers concerning to the portfolio

diversification can create larger risk premium to offset the firm-specific risk of

unaudited financial information. In summary, it is through audits that investors reduce

both market-related risk and firm-specific risk (Wallace 1980; Shakun, 1978).

According to Wallace (1980), the monitoring hypothesis seems to overlap with the

information hypothesis since the part of the audited information that is valuable to

agents and principals is also applicable to investors for their investment decisions.

However, he further suggests that the monitoring hypothesis provides support for the

practice of furnishing principals with an audited financial statement only within the

period of the contract agreement (i.e. within the duration of the agent-principal

relationship). According to the information hypothesis, financial information

determines market value. Investors require financial information in order to make a

rational investment decision even though they are on the outside of a contract of agent

and principal relationships. In other words, in order to make investment decisions,

investors need financial information from firms on a continuous basis and without

time limits

2.3.3 Signalling or reputation hypothesis

Wallace (1980: 30) notes that “signalling is a kind of implicit guarantee”. In an

agency relationship in which information asymmetry problems arise, the suppliers of

financial statements are assumed to be dishonest in reporting financial information.

As such, the users of financial statements are incapable of distinguishing between

honest and dishonest information. In this case, the demand for independent audits can

be seen to result in the financial statement users receiving honest reports (Wallace,

1980). Thus, audit services inform the market that the financial statements that are

provided by management are also free from material errors. Such assurance provides

the confidence to investors and other users of a financial statement that the reported

accounting numbers are reliable. As pointed out by Wallace (1980: 36), “specifically,

an audit can signal less noise or error in the financial report, greater fineness in the

reporting methods (including with GAAP), and unbiased performance measures.”

Furthermore, the signalling hypothesis offers an explanation for the demand for

different levels of audit quality. According to Moizer (1992), in a market where

sellers are unable to build a reputation, two major agency problems (moral hazard and

adverse selection problems) collaborate to diminish the quality of the product. If

buyers fail to make a distinction between the different levels of audit quality, they

may view all audit services as being of average quality and will only be willing to pay

for them at the same price. The audit providers do not, therefore, have any way of

influencing a buyer to acquire their services in preference to any others. As a result,

the moral hazard problem will arise because providers are likely to sell low quality

and low cost services in order to maximise their profits and the profits that would

come from providing good quality services are now accumulated among providers

regardless of the quality of an individual provider (Moizer, 1992).

Simultaneously, the adverse selection problem could also arise because of the

possibly that the market will become driven by low quality providers and good quality

providers will be forced to desist from the market (Moizer, 1992). The consequence of

these effect trades of average quality services is that the market becomes smaller, and

this leads to the potential for market collapse (Akerlof, 1970). The signalling

framework provides a cure for market collapse because it explains the sellers‟ ability

to provide a signal to uninformed buyers about the quality of their products or

services where there is an assumption that the seller knows the quality of their product

and the buyer does not (Bar-Yosef and Livnat, 1984).

Since buyers are unable to determine the quality of a product in advance, several

models of reputation capital suggest that the seller needs to expend resources in order

to establish a reputation (Klein and Leffler, 1981; Shapiro, 1983; Rogerson, 1983;

Allen, 1984). For example, Klein and Leffler (1981) argue that higher quality sellers

invest in non-salvageable firm-specific assets (e.g. advertising or marketing

investment) in order to prevent their competitors from entering the market and thus

they provide direct value to buyers. Shapiro (1983) suggests that sellers can establish

their reputation by initially charging for a higher quality product at a minimum quality

price that is equivalent to the cost of production because they are new entrants to the

market. In the early period the sellers may suffer economic losses, but later they

recover the price premium, provided that they maintain the production of higher

quality products. As pointed out by Shapiro (1983: 669):

“… the premium for a high quality product represents only a fair rate of

return on the investment in reputation. The typical time pattern of profits

to a seller is given by an initial period of losses, i.e., investment in

reputation, followed by a stream of profits… The higher the quality

produced, the larger are the initial losses (investment in reputation) and

the subsequent profits (premiums for high quality items).”

Allen (1984) disagrees with the models proposed by Klein and Leffler (1981) and

Shapiro (1983) by saying that investments in non-salvageable firm-specific assets are

not practical in some industries and that sellers should probably not charge for a

higher quality product at a minimum quality price and thus suffer losses in the initial

period of investment. Allen (1984: 312) argues that sellers that produce a higher

quality product should price it at a higher price which can be above the marginal cost.

He claims that buyers “reassure themselves about high quality of each firm‟s output

by verifying that the price charged and quantity produced are consistent with high

quality‟s being more profitable than low quality”. When a seller charges for a high

quality product at lower cost, it is perceived by the buyer that the seller has

transformed a higher quality product into a lower quality product, and this leads to the

buyer's resistance to purchasing any of the seller's outputs.

Once the sellers‟ reputations have been established, they are then able to signal to the

buyers that their products are endorsed with higher quality marks. Klein and Leffler

(1981) suggest that firms with an established reputation are less likely to produce a

low quality product because once the buyers are aware that they have purchased such

a product, this information will quickly be disseminated to other buyers. Once their

reputation is damaged, sellers may fail to secure an adequate return on their quality

product (Klein and Leffler, 1981; Shapiro, 1982; 1983; Rogerson, 1983).

Figure 2.2: The investment in reputation and the premium for high a quality

product (Source: Shapiro, 1981)

In relation to the audit market, Moizer (1997) claims that the signalling hypothesis

does not necessarily entail a higher quality audit because it simply leads the market

users to believe that the more expensive auditing firms offer a higher quality of

service. Consistent with this is DeAngelo's (1981) assertion that since audit quality is

unobservable and costly to measure, the market tends to use good reputation, derived

from large auditors, as a signal of a higher quality audit. As mentioned in Shapiro

(1983: 659):

“The idea of reputation only makes sense in an imperfect information

world. A firm has good reputation if consumers believe its products to be

of high quality. If product attributes were perfectly observable prior to

purchase, then previous production of high quality items would not enter

into consumers‟ evaluations of a firm‟s product quality. Instead, quality

beliefs could be derived solely from inspection.”

2.3.5 Insurance hypothesis

The insurance hypothesis differs from the agency relationship hypothesis and it

applies when auditors are involved in litigation. It suggests that audit services provide

investors with a form with of protection in the event of an audit failure (Wallace,

1980; Menon and William, 1994; Stice, 1991). In other words, the legal system allows

the investors to recover their investment losses from the auditor if the audited

financial statements contain a misrepresentation or a low quality audit. The

probability of recovering such claims increases if the auditors are among the larger

audit firms or those known as „deeper pockets‟ (Schwartz and Menon, 1985).

Wallace (1980: 21-22) provides four explanations of why managers choose auditors

as insurance in preference to insurance companies. Firstly, society assumes that

managers who fail to guarantee that they are fully independent of their actions,

without the auditors‟ attestation, are committing fraud or are involved in negligence.

Secondly, the improvement in accounting and auditing firms that employ legal staff,

legal services and in-house counsels, suggests that they are more efficient compared

with insurance companies. Thirdly, the insurance companies use the cost-benefit

approach when deciding whether to enter a legal defence or to decide on an out of

court settlement. However, both the auditors and the firms that are involved in

litigation will consider the effect on their reputation and thus, with a similar common

interest, they will ensure that they protect their reputations. Fourthly, if investors

suffer from losses because of an audited financial statement, the courts are likely to

hold the auditors responsible and to require them to bear the losses. An auditor‟s

contributions to an investor‟s losses are viewed by the court as „socializing risk‟. As

stated in Wallace (1980: 22), “… because he is responsible for business failures, the

auditor in turn shifts this cost to clients through higher fees and then to society

through higher prices and lower returns on investment.”

Several studies have empirically tested the insurance hypothesis. For instance, Menon

and Williams (1994) use the case study of Laventhol and Horwath (L&H)14

and they

examine the effect of L&H clients‟ stock price (1) when L&H filed their bankruptcy

Laventhol and Horwarth (L&H) is one of the seventh-largest public accounting firms in the U.S.

They declared bankruptcy on November, 1990 and filed for protection under Chapter 11 of the US

Federal Bankruptcy Code.

and (2) on the announcement of a replacement auditor. They hypothesize that when

L&H filed for bankruptcy, they came to the end of the operation and their investors no

longer had access to recover their investment losses. Thus, their clients‟ stock prices

were expected to decline. When the clients of L&H reappointed a new auditor, they

assumed that the investment losses from L&H were not transferable to the new

auditor, since investors can only claim from them if they have used the audited

financial statement (prepared by the new auditor) for their investment decision. If the

insurance hypothesis bears upon the L&H bankruptcy case, such a new appointment

could provide a significant reaction in stock prices. However, if the market perceives

that the appointment of new auditors could clarify the uncertainty of future

monitoring, it may result in positive return. Their findings are consistent with the

insurance hypothesis. The price reaction on both events supports the argument for the

absence of the expected insurance coverage thus: the disclosure of the L&H

bankruptcy had negative impact on their stock prices and the announcement of a

replacement new auditor did not provide any significant reaction.

In addition to this, Baber et al (1995) suggest that such price reactions were also

driven by the monitoring function of L&H in which the insurance and monitoring

hypotheses are difficult to differentiate. They suggest that financially distressed

auditors are more likely to perform low quality audits because they are more

concerned with their current position than with their competence and independent

judgement. For example, in order to retain their clients and minimise their audit cost,

financially distressed auditors are less likely to report an error or a misstatement that

they discover during the auditing work or they may reduce audit testing in order to cut

down the audit cost. They argue that if the investors were aware and if they perceived

that L&H was incompetent and independent, then such a perception may have forced

the stock prices to decline. However, Lai and Gul (2008) provide contradictory

evidence to Baber et al. (1995). Using the likelihood of issuing modified audit

opinion, the provision of discretionary accruals and the predictability of discretionary

accruals for future earnings as proxies for audit quality, they suggest that the audit

quality of L&H was not substandard.

In another study, O‟reilly et al. (2006) examine the interaction of signalling and

insurance hypotheses through studying audit opinion in an experimental setting. They

argue that a going-concern audit opinion (1) provides signals to the market that the

firm is no longer feasible, thus affecting stock prices; (2) offers the auditor legal

protection, although there is a possibility that investors are able to recover part of their

investment‟ losses and (3) increases the value to investors because of the increased

need for insurance coverage. Their findings suggest that the going-concern audit

opinion reduces analysts‟ estimation of stock price because market participants

consider the auditors‟ role as an insurance protector. In sum, the insurance hypothesis

provides the most consistent support for the view that auditors are perceived by

investors as the guarantors of their investment and investors “appear to be willing to

pay a premium for the right to recover potential investment losses from auditors

through litigation” (Menon and William, 1994: 341).

As well as increasing the direct costs that auditors need to charge for in order to cover

for investors‟ losses, such lawsuits also have an indirect impact on their reputation

and perceived audit quality (Palmrose, 1991). The findings of Chaney and Philipich

(2002) are consistent with those of Menon and Williams (1994) and Baber et al.

(1995). They investigate the impact of the Enron audit failure on Arthur Andersen's

(A&A) reputation as one of the big-five auditors. They examine the A&A clients‟

stock prices on the three days after A&A admit they shredded a significant number of

audit documents related to the Enron engagement. Such an unexpected event results in

a significant negative market reaction on the A&A clients‟ stock prices, suggesting

that investors acted on the perceived low quality of the audit performed by A&A

In a similar form of study Hillison et al. (2004) examine the audit clients‟ stock price

reaction related to Ernst & Young‟s (EY) rumours of bankruptcy in late November

and early December 1990 (event studies). Their findings suggest that the insurance

hypothesis and the audit quality explanation account for the negative stock price

reaction. Even though the big-4 auditors may provide a higher quality audit, market

participants still react according to newly published information. When market

participants lose their confidence in the credibility of audited financial statements, it

effects a reduction in clients‟ stock price.

Lennox (1999) has empirically tested the insurance hypothesis and reputation

exposure (under the signalling hypothesis) using UK data between the periods of 1987

and 1994. According to the reputation hypothesis, the big-size auditors signal their

audit quality by assuming that they are more likely to lose their client-specific rent

when they provide a low quality audit. In order to avoid such loss, they have more

incentive to provide a higher audit quality (DeAngelo, 1981). The alternative to this

argument is that the wealth auditors or big-size auditors are associated with higher

litigation risk (Dye, 1993). Similarly, in order to prevent such larger litigation claims

(e.g. because of low quality audit) from the investors, the big-size auditors offer a

more credible and higher audit quality. Lennox (1999) posits that the lower the audit

quality undertaken by the auditors, the higher the potential of such auditors to be sued

(because they fail to detect and report misstatement or negligence). He argues that, in

the case of the big-size auditors that gain their „quality‟ from reputation capital, the

auditor‟s litigation provides an accurate indicator. However, the insurance hypothesis

considers the auditor‟s litigation to be a poor indicator because the auditors are likely

to be sued if they are insufficiently conservative (type I error), but they will not be

sued are if they are too conservative (type II error). Thus, although the big-size

auditors provide higher audit quality than a smaller size auditor, there is a higher

probability that they will be sued when a type I error arises. Lennox‟s findings suggest

that the big-size auditors are more likely to be sued because they are more fearful of a

potential litigation claim than of losing their client-specific rent or reputation capital.

2.4 The association between board of directors, audit committee, external

auditor and financial reporting quality

There are two research questions to be examined in this thesis. The first question

concerns the relationship between the effective board and audit committee, and audit

quality. The second question is concerned with the relationship between the effective

board of directors, audit committee and auditor quality in constraining earnings

management. Prior to explaining the reasons why these parties demand a higher

quality audit and why they are more likely to constrain earnings management, it first

reviews the roles of boards of directors and audit committees and their connection to

an external audit.

The board of directors is the main player in the success of a firm. They are responsible

for setting the goals, strategies and values of a firm, in order to align them with the

interests of their shareholders (UK Corporate Governance Code, 2010). They are also

responsible for the transparency and fairness of financial statements, a duty which has

been clearly stated in the Companies Act 1985 and the Company Reform Bill.

Section 226 of the Companies Act 1985 requires directors to prepare and assume

responsibility for the individual account and administration of the firm. While,

Section 366 (1) of the Company Law Reform Bill (2005: 164) states that directors

must not approve these accounts except they are satisfied themselves that the accounts

have give true and fair view and have been prepared accordingly to the relevant

financial reporting framework. These accounts, having been prepared and approved

by the directors, are required to be audited by an external auditor as they are to be

used by the public.15

As the highest point in the hierarchy of a firm's structure, the

board of directors is accountable for all a firm‟s activities, strategies and financial

performance including the sub committees‟ actions.

Under a main board, there are several subcommittees, one of which is the audit

committee. The audit committee has a direct link with the financial performance of

the firm and the external audit services. Wolnizer (1995: 47-48) discusses in detail the

tasks that audit committee members are expected to perform, from three perspectives:

(1) accounting and financial reporting – The audit committee reviews financial

statements, accounting policies, fraud, internal control, changes and all

significant matters that could affect the financial statements.

(2) auditors and auditing – The audit committee provides recommendations for

the external auditors, reviews the scale of audit fees and non-audit fees, writes

a letter of engagement, reviews the audit plan, ensures auditor independence

and allocates resources on the internal audit. In relation to the audit fees,

Collier and Gregory (1996) claim that an effective audit committee is

responsible for ensuring that the scope of an audit is sufficient, and to some

extent the audit committee is able to ensure that the reduction of audit fees

does not reach a level where it may potentially jeopardize the quality of audit

As stated under Section 235 and 237 of Company Act, 1985 and Section 485to 488 of Company

Reform Bill.

(3) corporate governance – The audit committee facilitates the relationship

between the auditors and the board of directors, as well as reviewing and

complying with corporate policies, ethical policies and codes of conduct.

By implementing these tasks (1 to 3), firms are expected to enhance the credibility,

objectivity and reliability of financial statements, improve the accountability of

management staff, reduce any opportunistic behaviour of management, increase the

efficiency and effectiveness of internal control as well as that of internal and external

audits and strengthen the function of the board of directors while helping them to

meet their legal responsibilities (Wolnizer, 1995: 49). The overall implication is that

the activities of an audit committee could improve the quality of financial reporting

and corporate governance of firm.

Similarly, Menon and Williams (1994: 123) suggest two possible advantages to be

gained by establishing an audit committee. Firstly, the independent audit committee

may act as an independent party between the internal audit and the external audit. The

independent members of an audit committee help them to provide an unbiased

assessment between internal audit function and external audit services, which in turn

improves the quality and integrity of a firm‟s financial statements (Imhoff, 2003).

Secondly, the audit committee may enhance the efficiency of a board‟s function

particularly when the board has a large number of directors

Furthermore, The Blue Ribbon Committee (BRC, 1999: 19) also agrees that the

formation of an audit committee can enhance the credibility of the reported financial

statements and thus maintain the investors‟ confidence. They stated that:

“…the Committee believes audit committee will be more effective in

helping to ensure the transparency and integrity of financial reporting and,

thereby, maintain the investor confidence that makes our securities

markets the deepest and most liquid in the world.”

In addition to the important role of a board of directors and an audit committee,

Bailey and Grambling (2005) suggest that an external audit serves as a key

determinant of financial reporting quality. Power (1996) claims that an external audit

adds credibility to the financial report provided by management. DeAngelo (1981)

argues that the auditors improve the quality of financial reports through their

competency and independence audit. Furthermore, Ruddock et al. (2006) claim that

the quality of financial reporting is improved when the auditors respond to the

aggressive earnings conservatism.

Anderson et al. (2001) posit that when a manager has a higher incentive to manage

earnings, the auditor perceives that manager as more aggressive, as having a greater

desire to look good in their financial statements and as expecting the auditors to agree

with their financial statements. Therefore, auditors will limit earnings management

when they believe that managers are manipulating financial statements. Moreover,

according to Krishnan (2003b), by constraining earnings management, the auditors

are able to improve the information value of earnings. If the market perceives that

auditors are unable to limit opportunistic earnings, then the earnings‟ information

value would be diminished simultaneously. According to Sankar and Subramanyam

(2001), the restriction imposed by GAAP, and by auditors, on the discretion of

reporting earnings may improve the content of earnings‟ information.

There is a direct and an indirect link between the role of a board, an audit committee

and an external auditor. Under the direct relationship, the principal roles of an audit

committee are to make a recommendation to the board in relation to the appointment

of the external auditor, to review the audit engagement and the audit fees and to

monitor the external auditor‟s independence and objectivity as well as the

effectiveness of the audit process (UK Corporate Governance Code, 2010). As

previously stated, in relation to the audit engagement and proposed audit fees, an

effective audit committee is responsible to ensure that the scope of an audit is

sufficient and that the proposed audit fees do not potentially jeopardize the quality of

auditing work (Collier and Gregory, 1996). The reason for this is because auditors

seek to minimise the total cost of an audit and endeavour to achieve a balance

between the costs of audit resources and expected future losses as a result of legal

liability (Carcello et al., 2002). It is reasonable to expect that an effective board of

directors first reviews the overall scope of an audit and the proposed audit fees before

accepting a proposal from the audit committee, since the board of directors is

accountable for all their the sub committees‟ actions. In respect of auditor

independence, specifically the provision of NAS, the authoritative guidance requires

an audit committee to review NAS engagement and make sure that relevant

procedures are in place to ensure that the independence and objectivity of the auditors

are not affected by the undertaken NAS. An audit committee is responsible for

reporting and making recommendations to the board of directors on any actions taken

to ensure that the auditor's independence has been safeguarded (UK Corporate

Governance Code, 2010: 20).

Indirectly, an effective board and audit committee may signal to management and

auditor that they exercise a higher and more vigilant oversight function. For example,

when management perceives that a board and audit committee are performing a

higher monitoring function, they may voluntarily consider limiting the purchase of

NAS (Abbott et al., 2003b) and will limit their own opportunistic earning behaviour

by employing higher quality auditors. Similarly, auditors may see that an effective

board and audit committee are associated with having a higher monitoring function

and they are thus likely to be more demanding and to insist on having a higher quality

audit (Carcello et al., 2002).

Why do boards of directors and audit committees demand different levels of audit

quality? Why do they constrain earnings management? Similarly, why do external

auditors limit opportunistic earnings management? The answers to these questions lie

in their implications for reputation capital, legal exposure and shareholder interests

(Carcello et al., 2002).

The reputation hypothesis posits that vigilant directors make costly investments to

establish their reputation as effective monitors and, in return for being good monitors,

they might be rewarded with an additional directorship in another firm (Fama, 1980;

Fama and Jensen, 1983). Evidence suggests that when directors suffer a damaged

reputation, they are less likely to get opportunities to serve on another board. For

example, Gilson (1990) claims that the outside directors of firms in financial distress

hold significantly fewer seats on other boards following their departure, possibly due

to reputation effect and legal exposure. In another study, Fich and Shivdasani (2007)

examine the reputation impact for the outside directors of firms that are involved in

financial fraud. They find that outside directors lose about 50 percent of their

directorships in other firms when one of the firms that they serve is involved in

financial fraud lawsuits. This finding indicates that sued directors on a board are seen

as weak monitors who may increase the likelihood of financial misconduct occurring.

Moreover, they also find that a reduction in directorships may be driven by an outside

directors‟ desire to reduce their future legal exposure.

The reputation hypothesis is also applicable to auditors. A highly reputable auditor

has an incentive not to produce a low quality audit because, once their clients discover

that they provide a low quality audit, their reputation will be damaged and they will

be unable to secure their clients and they will lose quasi rents (DeAngelo, 1981).

Wilson and Grimlund (1990) provide evidence of the consequences that auditors may

suffer if their reputations are damaged. They examine the effect of SEC disciplinary

actions on audit firms and their findings suggest that the auditors tend to lose their

market share and that they experience difficulty in retaining clients. In general, the

auditors are likely to constrain earnings management because of the possibility of

being sued or subjected to regulatory actions. These may be due to negligence in

identifying misleading information in the audited financial statement. Evidence

suggests that auditor litigation has positive relationship with earnings management

(Lys and Watts, 1994; Heninger, 2001)16

and a failure to fulfil their role effectively or

a neglect of their duties may increase an auditor's potential for future legal exposure

(Lennox, 1999).

As well as considering reputation capital and legal exposure, a board of directors and

audit committee demand a higher quality audit in order to promote the shareholder's

interests (Carcello et al., 2002). Several studies have suggested that investors perceive

that the provision of NAS negatively affects auditor independence and undermines the

audited financial statement. Lavin (1976; 1977) and Firth (1980) examine the

perception of accountants, financial analysts and loan officers in the US and UK,

respectively. Their findings suggest that when auditor independence is deemed to be

impaired, investment and borrowing decisions will also be affected. These studies

may suggest that investors are unfavourable towards a firm if they perceive that the

auditing of financial statements has been impaired by the purchase of NAS.

Therefore, the board of directors and audit committee monitor auditor independence

The auditors are subject to lawsuits for the overstated earnings (Lys and Watts, 1994; Heninger,

2001) but reveal no evidence for understated earnings (St. Pierrie and Anderson, 1984).

(e.g. in terms of the levels of NAS) in order to gain the investors‟ confidence and

promote the shareholders' interests.

A higher quality auditor may perceive by the investors to be associated with a higher

credibility of information, which in turn increase the firm value (Titman and

Trueman, 1986; Datar et al., 1991). Indeed the investors assume that the higher

quality auditors are more sensitive to earnings surprises. For example, the existing

studies suggest that the firms engage or switch to big-size auditors have lower

earnings management and higher earnings respond coefficient compared to smaller

size auditors, consistent to the view that big-size auditors provide more credible

information to investors (Teoh and Wong, 1993; Becker et al., 1998). In other study,

Khurana and Raman (2006) suggest that the higher NAS fees and total fees received

by the auditors are viewed negatively by the investors because higher fees could

possibly compromise the auditors‟ independent and audit quality through the auditors‟

economic bonding on clients. Such views have been expressed into the investor

perception as a lower ex ante cost of equity capital.

Overall, higher quality audits and a higher quality of reported earnings are not only

useful for investors and the users of financial statements, but they are also beneficial

to auditors, boards of directors and audit committees because they are able to

minimise the potential for damaged reputation and legal exposure while also raising

the support of shareholders.

2.5 Summary

Agency theory assumes that principals and agents have divergent interests and thus

are likely to contribute to agency conflicts that include the phenomenon of earnings

management. To align these interests, agency theory recognises the monitoring roles

of a board of directors, an audit committee and external auditing as playing a part in

the mitigation of agent-principal conflict. From the agency perspective there are

several characteristics of board and audit committee (e.g. size, composition, expertise

and levels of activities) that contribute to an effective monitoring function. The

independent audit is also acknowledged by the agency theory as a control mechanism

to reduce information asymmetry between the shareholders/investors and

management by promoting truthfulness and fairness in financial statements. Several

hypotheses have explained why shareholders/investors or management demand

auditing services and different levels of audit quality. These hypotheses include the

monitoring hypothesis, information hypothesis, signalling/reputation hypothesis and

insurance hypothesis. By employing a higher quality auditor and constraining

earnings manipulation, a board of directors and audit committee assume that they are

adding credibility to the financial statement and increasing a firms‟ value. At the same

time, the board and audit committee are able to secure their reputation capital, avoid

legal exposure and promote the shareholders' interests. Similarly, a higher quality

auditor is less flexible towards opportunistic earnings due to the risk that wrongly

reported earnings may incur reputation damage, increase future legal exposure,

decrease a firm‟s value and disappoint shareholders.

CHAPTER 3

LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT

3.1 Introduction

This chapter reviews the existing literature on three topics: audit quality, corporate

governance and earnings management. It first reviews the definition of audit quality

and how it is measured, and this is followed by discussion and review of the effective

characteristics of boards and audit committees. Previous studies of earnings

management, particularly those related to the motivation for earnings and earnings

management measurement, are also reviewed. These reviews provide a general

understanding of the areas of study that is being investigated in this thesis.

Towards the end of this chapter, the discussions and reviews focus on the association

between corporate governance and audit quality and between corporate governance

and auditor quality in respect of constraining earnings management. These reviews

help to identify similar studies that have been done and which provide potential

evidence of research gaps that demand further investigation. For each of the main

relationships, the development of tested hypotheses is also disclosed. Finally, the

summary and conclusion are presented in the last section.

3.2 Definition of audit quality

The most common definition for audit quality is derived from DeAngelo (1981: 186),

she defines audit quality as “the market assessed joint probability that a given auditor

will both (a) discover a breach in the client‟s accounting system and (b) report the

breach.” This definition is accessed by market as the ability of an auditor to detect

accounting misstatements and then to express them in appropriate audit opinion.

Watts and Zimmerman (1986) simplify DeAngelo‟s definition where the first part

refers to auditor‟s competence and the quantity of inputs devoted to the audit, while

the second part refers to an auditor‟s independence. In other words, according to

Watts and Zimmerman, any factors that are associated with a lack of auditor

competence or a lack of auditor independence are able to compromise the quality of

audit. 17

Palmrose (1988) describes audit quality in terms of levels of assurances. Higher levels

of assurances (i.e. possibility that financial statements contain fewer errors or

misstatements) are associated with a higher audit quality and vice-versa. The grounds

for this definition have been developed from audit failures (where an auditor has

failed to detect a material error or misstatement) than can be traced in litigation cases.

According to Francis (2004), audit failure can be classified as extremely low audit

quality (end quality) that can result in several outcomes such as regulatory sanctions,

litigation rates, business failure and earnings restatement.

From the regulator‟s perspective, ICAEW (2002: 8) suggests a definition for audit

quality by stating that, “At its heart [audit quality] is about delivering an appropriate

professional opinion supported by the necessary evidence and objective judgements.”

As long as the auditors provide an independent audit opinion that is supported by

adequate audit evidence, the regulator assumes that such auditors have performed a

quality auditing service.

Despite the fact that technical qualities, such as an auditor‟s ability to detect and

report errors, have been argued as the defining aspects of audit quality, Duff (2004)

suggests that audit quality is made up of both technical quality and service quality (the

levels of clients‟ satisfaction and expectations). Technical quality consists of

reputation capital, capability, expertise, experience and independence scales, while

Beattie et al. (1999: 71) claim that there are two broad categories of factors that affect an auditor's

independence. They are, namely, economic and regulatory factors. They suggest that the economic

factors are related to an auditor‟s economic dependence on clients, competition among auditors in the

audit market, the provision of NAS, and deficiencies in the regulatory frameworks. The regulatory

factors are those factors that are associated with accounting and auditing standards. For example in

promoting the auditor independence, the Cadbury report (1992) provides the recommendation that the

formation an audit committee is able to strengthen the auditors‟ independence and thus to enhance

audit quality. Arruñada (2000) appends that the term “independent” is not exclusively confined to the

relationship between the main client and auditor, but that it also encompasses other important attributes

such as third parties and other clients. He cites the example of a client being more worried when the

company of an auditor‟s other client collapses without prior notice than they are about having low

quality audit. As a consequence, “clients have very strong incentives to monitor, evaluate and

compensate audit firms‟ quality” (Arruñada 2000: 218).

service quality is defined by responsiveness, empathy and the provision of NAS and

client services.18

Following DeAngelo‟s and Watt and Zimmerman‟s definitions, this thesis defines

audit quality as the competence of the auditors to detect errors and the objectivity (in

fact and in appearance) of the auditors in reporting such errors.19

The terms “audit

quality” and “auditor quality” are assumed to be synonymous, and this is consistent

with Clarkson and Simunic‟s (1994: 208) suggestion that “an audit firm is assumed to

supply a single level of audit quality that at a moment in time.”

Several factors may influence the quality of an audit. Wooten (2003) claims that audit

firms, audit teams and the professional judgement or auditor independence are the

principal contributors to audit quality (as shown in Figures 3.1). A audit firm and

audit team factors (e.g. human resources, audit processes, industry expertise,

supervision, audit planning, and professionalism) directly contribute to the skill and

competence of auditors in detecting errors and misstatements. The factors of audit

tenure, audit fees and NAS not only directly impair auditor independence, but they

also implicitly support auditor effectiveness.

In addition, FRC (2008) suggests five key drivers for audit quality: (1) the audit firm

culture, (2) skills and personal qualities of audit partners and staff, (3) the audit

process, (4) usefulness of the audit reporting and (5) factors that are outside the

control of the auditors (refer to Figure 3.2 for detailed outlines of key drivers). FRC

(2008) suggest that the roles of internal governance mechanisms (e.g. audit

committees) and regulatory requirements may help to improve audit quality. An

The independence and NAS scales are relatively different from each other (Duff, 2004: 110).

19 Consistent with AICPA (2009), in this thesis, the auditor objectivity or auditor independence in fact

(mind) and in appearance are defined as: “Independence of mind—The state of mind that permits the

performance of an attest service without being affected by influences that compromise professional

judgment, thereby allowing an individual to act with integrity and exercise objectivity and professional

scepticism. Independence in appearance—The avoidance of circumstances that would cause a

reasonable and informed third party, having knowledge of all relevant information, including

safeguards applied, to reasonably conclude that the integrity, objectivity, or professional scepticism of a

firm or a member of the attest engagement team had been compromised.”

effective audit committee is capable of enhancing audit quality through active

involvement during the audit period and effective communication with the auditors.

Figure 3.1: A model of audit quality (source: Wooten 2003)

To sum up, audit quality can be described as the ability of an auditor to provide an

independent audit which results in a financial statement that is free from

misstatement, error and fraud. Since an audit‟s quality is influenced by three main

parties (audit firm, audit‟ client and regulators), the attributes or factors that are

associated with each group can be used as indicators for audit quality.

Drivers Indicators

The culture

within an

audit firm

The culture of an audit firm is likely to provide a positive contribution

to audit quality where the leadership of an audit firm:

Creates an environment where achieving high quality is valued,

invested in and rewarded.

Emphasis the importance of „doing the right thing‟ in the public

interest and the effect of doing so on the reputation of both the

firm and individual auditors.

Ensures partners and staff have sufficient time and resources to

deal with difficult issues as they arise.

Ensures financial considerations do not drive actions and decisions

having a negative effect on audit quality.

Promotes the merits of consultation on difficult issues and

supporting partners in the exercise of their personal judgment.

Ensures robust systems for client acceptance and continuation.

Fosters appraisal and reward systems for partners and staff that

promote the personal characteristics essential to quality auditing.

Ensures audit quality is monitored within firms and across

international networks and appropriate consequential action is

taken.

The skills

and personal

qualities

of audit

partners and

The skills and personal qualities of audit partners and staff are likely

to make a positive contribution to audit quality where:

Partners and staff understand their clients‟ business and adhere to

the principles underlying auditing and ethical standards.

Partners and staff exhibit professional scepticism in their work and

are robust in dealing with issues identified during the audit.

Staff performing detailed „on-site‟ audit work has sufficient

experience and are appropriately supervised by partners and

managers.

Partners and managers provide junior staff with appropriate

„mentoring‟ and „on the job‟ training.

Sufficient training is given to audit personnel in audit, accounting

and industry specialist issues.

continued

Drivers Indicators

effectiveness

of the audit

process

An audit process is likely to provide a positive contribution to audit

quality

where:

The audit methodology and tools applied to the audit are well

structured and:

o Encourage partners and managers to be actively involved in

audit planning.

o Provide a framework and procedures to obtain sufficient

appropriate audit evidence effectively and efficiently.

o Require appropriate audit documentation.

o Provide for compliance with auditing standards without

inhibiting the exercise of judgment.

o Ensure there is effective review of audit work.

o Audit quality control procedures are effective, understood and

applied.

High quality technical support is available when the audit team

requires it or encounters a situation it is not familiar with.

The objectives of ethical standards are achieved, providing

confidence in the integrity, objectivity and independence of the

auditor.

The collection of sufficient audit evidence is not inappropriately

constrained by financial pressures.

reliability

usefulness of

reporting

Audit reporting is likely to provide a positive contribution to audit

quality where:

Audit reports are written in a manner that conveys clearly and

unambiguously the auditor‟s opinion on the financial statements

and that addresses the needs of users of financial statements in the

context of applicable law and regulations.

Auditors properly conclude as to the truth and fairness of the

financial statements.

Communications with the audit committee include discussions

about:

o The scope of the audit.

o The threats to auditor objectivity.

o The key risks identified and judgments made in reaching the

audit opinion.

o The qualitative aspects of the entity‟s accounting and

reporting and potential ways of improving financial reporting.

continued

Drivers Indicators

Factors

outside the

control of

auditors

Factors outside the control of auditors which are likely to make a

positive contribution to audit quality include:

An approach to corporate governance within the reporting entity

that attaches importance to corporate and financial reporting and to

the audit process.

Audit committees that are active, professional and robust in

dealing with issues identified during the audit.

Shareholders that support auditors, where appropriate, thereby

increasing the likelihood that directors and management will

comply with their obligations in relation to the preparation of

reliable financial statements.

Reporting deadlines that allow the opportunity to carry out an audit

without undue reliance on work performed before the end of the

reporting period.

Appropriate agreed arrangements for any limitation of liability.

An audit regulatory environment that focuses on the drivers of

audit quality.

Figure 3.2: The key drivers of audit quality (source: FRC, 2008)

3.3 A possible approach to measuring audit quality

The measurement of audit quality is complex and problematic (see e.g. Wooten, 2003;

Niemi, 2004; Jensen and Payne, 2005). However, Bailey and Grambling (2005),

Francis (2004) and PCOB (2008) have identified several potential measures for audit

quality in both academic literature and in practice. These measures are perceived as

factors, indicators, behaviors or perceptions that have a direct and an indirect link with

audit quality as defined in the previous section.

Bailey and Grambling (2005) and PCOB (2008) discuss potential measures of audit

quality based on the audit process that is adhered to in completing an audit

engagement. These measures are associated with the audit procedures, the

documentation of audit evidence and compliance with auditing standards. They

classify them as input and output-based measures. Bailey and Grambling (2005)

suggest that the inputs of audit processes are relevant to the quality control system of

an audit firm. These include (1) how audit firms put an effort into promoting and

emphasising desirable qualities (e.g. independence, objectivity, and professionalism),

(2) internal control (e.g. an audit firm‟s internal review), (3) human resources (e.g.

competence and independence of staff) and (4) audit methodologies (e.g. audit

policies and procedures). Moreover, in respect of human resources, Bailey and

Grambling (2005: 15) suggest that the skills and competencies of auditors should be

viewed in a broader context which goes beyond technical accounting and auditing

skills. They argue that the level of professional scepticism of auditors may affect their

ability to act independently when executing auditing work. Thus, the attribute of

„independence‟ is desirable for every auditor and audit team member when it comes to

achieving a higher audit quality.20

In short, Bailey and Grambling (2005: 14) state that

“if audit quality is defined on the basis of inputs, those inputs could be loosely

described as the „right‟ people, applying the „right‟ tools and procedures in the „right‟

organisational culture which includes appropriate internal monitoring”. The output-

based measures specifically relate to the audit opinion and whether that audit opinion

reflects the “accuracy of management‟s assertions”, and it includes the issuance of

going concern audit opinion and restatements (Bailey and Grambling, 2005:13).

As an alternative to the suggestions that are given by Bailey and Grambling (2005),

PCOB (2008) defines the input-based measures as the things and procedures that

auditors have taken into account in arriving at a given audit opinion (e.g. the audit

procedures used in detecting fraud, the experience levels of audit staff and audit

partners, and the annual staff retention). The output-based measures refer to outcome

or evidences that an auditor produces from work that has been undertaken. Such

outcomes, for instance, can be measured by the number of frauds discovered and

numbers of errors or misstatements detected.

In general, the potential measures of input and output given by Bailey and Grambling

(2005) and PCOB (2008) are limited to the factors that are associates with the audit

process that was adhered to in completing an audit engagement. These factors of audit

quality are beyond the audit process itself. A users‟ perception of audit quality is

claimed to be one of the alternative measure for audit quality (e.g. audit fees, NAS

fees and industry specialist auditor). Specifically, users reckon audit quality based on

an auditor‟s reputation. Khurana and Raman (2006) claim that a user‟s perception of

“Independence in fact allows the auditor to exercise judgement with an objective viewpoint, thus

achieving actual audit quality, while independence in appearance leads investors to trust that the

auditor is exercising professional judgement in an objective fashion” (Bailey and Grambling, 2005:15)

audit quality is important because it reflects public trust and confidence in a firms‟

reported financial information. Recognised the important these, in this thesis the audit

quality measures will be based on the users‟ perceptions.

Despite the various measures of audit quality that been used in the existing studies,

the present study acknowledges the limitation of these measures. For example, with

regard to an input-based measure, how do we ensure that the consistency of the

input‟s attributes has not been diminished during the process of the audit engagement?

Perhaps, information about the key drivers of audit quality, such as the education,

experience and competency of the auditors is not publicly available and difficult to

obtain. By using output-based measures the result of an audit is not necessarily

observable just after work has been undertaken because the quality of audit

information usually emerges over a certain period during which restatement or

business failure or the identification of misstatements can occur (PCOB, 2008:9). The

measures, such as audit firm size, that reflect an auditor‟s reputation may not be

essentially accurate measures of audit quality. While there are factors that are believed

to compromise auditor independence, such as NAS and audit tenure, it has also been

argued that these same factors can enhance an auditor‟s knowledge and competency.

The existing studies recognise that, to the certain extent, some measures (e.g. industry

specialist auditors) have demonstrated a strong relationship with higher audit quality.

The lack of the empirical evidence to support some measure as better audit quality

indicators, does not mean that these measures are insignificant, as they may

complement qualitative research on audit quality.

In this thesis, three measures of audit quality will be employed based on auditor

reputation and auditor independence points of view, namely, audit fees, NAS fees and

industry specialist auditors. These measures have been extensively used in prior

auditing research and each of them is now reviewed.

3.3.1 Audit fees

The link between audit fees and audit quality is suggested by the signalling or

reputation hypothesis (Lindberg, 2001). Models of reputation capital suggest that

sellers expend resources in order to build a reputation because buyers are unable to

determine product quality before purchasing (Klein and Leffler, 1981; Shapiro, 1983;

Rogerson, 1983; Allen, 1984). Once a sellers‟ reputation has been established, they

are able to convince buyers that their products have higher quality-assuring price, and

thus earn economic rents. If buyers discover that sellers produce a low quality product

(i.e. at the minimum-quality price), their established reputation will be impaired and

the economic rents assigned to the higher quality product will disappear (Klein and

Leffer, 1981). As concluded by Klein and Leffer (1981: 634), “our analysis implies

that consumers can successfully use price as an indicator of quality”.

There are several arguments for the proposition of audit fees as proxy for audit

quality. Several studies suggest that higher audit fees are associated with higher audit

quality in order to compensate for the high-price of reputation capital (i.e. big-size),

auditors‟ industry specialization, as well as for increased audit effort (see Simunic,

1980; Palmrose, 1986a; Craswell et al., 1995; Ferguson and Stokes, 2002).

Craswell et al. (1995) argue that the reputation development of an auditors‟ brand

name and industry specialization consumes a higher cost and thus results in higher

audit fees. Evidence suggests that audit clients are willing to pay a fee premium on

these auditors‟ reputations in order to get a better quality of service. The brand name

auditors demonstrate the effect of a Big 8/6/4 premium that is justified for higher

reputation capital, and thus they convey a higher quality differentiation compared to

non-brand name auditors (Simunic, 1980; Palmrose 1986a).

According to Palmrose (1986a: 108), the Big 8 auditors charge higher audit fees for

two reasons: they indicate (1) higher audit quality or (2) monopoly pricing. After

substituting the audit fees variable for audit hours, her finding supports that the Big 8

auditors are consistent with higher audit quality providers. She indicates that “big

eight designation is a quality surrogate, in that increased hours by big eight auditors

would reflect greater productive activities (evidence acquisition) in providing higher

levels of assurance (higher quality) to clients”. As well as the brand name auditors,

prior studies provide further evidence of the relationship between audit fees premiums

and industry specialist auditors. Craswell et al. (1995) and Ferguson and Stokes

(2002) claim that the brand name industry specialist auditors earn additional fee

premiums over non-specialist brand name auditors, which indicates a higher audit

quality differentiation among them.

In other study, Wolinsky (1983) argue that the price may signal a differentiation in

levels of quality. Although sellers are potentially capable of producing the preferred

and various levels of quality, the higher quality products are more costly to produce.

DeAngelo (1981) claims that larger sized auditors or auditors that earn higher fees

have more resources to invest when compared with smaller size auditors. Hence, they

contribute more to improving the quality of their work (DeAngelo, 1981).

Elitzur and Falk (1996) suggest that audit fees have a positive relationship with

planned audit quality. They examine planned audit quality and audit fees in a

multiperiod model. Ordinarily, higher audit fees might inspire auditors to increase the

audit quality. Hoitash et al. (2007) also agree that higher audit fees will increase the

auditor‟s effort and result in a higher audit quality.

In recent studies related to corporate governance, evidence suggests that lower audit

fees could also be associated with a perceived higher audit quality. This is because the

auditor might take into consideration that firms bound by a strong internal control

environment will probably have a lower audit risk thus reducing the audit effort and

audit fees by means of an effective internal corporate governance mechanism (Tsui et

al., 2001; Boo and Sharma, 2008b; Griffin et al., 2008; Krishnan and Visvanathan,

2009). Cohen et al., (2002) claim that an effective internal corporate governance

mechanism may potentially contribute to a higher audit quality by lessening audit

testing and enhancing the integrity of financial statements. Yeoh and Jubb (2001)

suggest that an auditor and the internal corporate governance mechanisms share an

identical common factor that contributes to a higher quality audit (i.e. independence

from management). Griffin et al. (2008) provide evidence that the demand for audit

services and internal corporate governance are co-determined by two countervailing

relationships. In the first relationship, an increase in audit fees can occur due to the

demand for effective corporate governance. The second relationship is associated with

a decrease in audit fees because the auditors benefit from strong corporate governance

and thus the audit risk and cost of auditing is reduced. Both relationships contribute to

a higher audit quality. Consistent with this, Tsui et al. (2001) and Krishnan and

Visvanathan (2009) suggest that those firms which separate the dual roles of CEO and

chairman and the audit committee equipped with financial expertise, are perceived by

auditors to have strong a internal control environment. This reduces the control risk

and audit effort, leading to lower audit fees.

Besides the empirical evidence on the relationship between audit fees and audit

quality that has been discussed above, several reports from the regulators emphasise

the importance of audit fee setting and how it may affect the quality of auditing work

(The Cohen Commission, AICPA 1978; Treadway Commission 1987; Cadbury

Report 1992; Advisory Panel on Auditor Independence, 1994). For example, The

Cohen Commission (AICPA, 1978) suggests that audit firms need to identify and

manage audit fees and other issues related to the audit resources (e.g. staff, time and

partners participation) as these factors are likely to devalue audit quality due to the

higher competition in audit market. Similarly, the authoritative body in UK also draws

attention to the audit fees factor and how it may influence audit quality. As postulated

by ICAEW (2002: 9) “audit quality is achieved only if it is the keystone of the firm‟s

overall strategy. Every single strategic decision taken by the firm will ultimately

impact on quality including the firm‟s policy on audit fess”. Concisely, ICAEW

alleges that an audit firm‟s policy on audit fees is one of the components that may

affect the quality of audits.

In opposition to the benefits of using audit fees as a proxy for audit quality, the

present study notices the limitation that audit fees are an imperfect measure of audit

quality. The audit fee is not necessarily accurate an as an indicator for audit effort as

the appropriate measure for audit effort is the number of audit hours. However, Deis

and Giroux (1996) provide some empirical evidence that audit fees and audit hour are

significantly related to audit quality in their analysis of three important attributes:

audit fees, audit hours and audit quality. Hence, it seems reasonable that more audit

hours will lead to higher audit fees and promote a higher quality audit. In addition, to

consider higher audit fees as a proxy for higher audit quality is consistent with

signalling or reputation hypothesis.

3.3.2 NAS fees

The issue of the provision of NAS has been discussed broadly in the prior literature

(see for example Beck et al., 1988a; 1988b; Beattie and Fearnley, 2002). The central

arguments about NAS are generally concerned with auditor independence and

whether the joint provision of audit and NAS enhances or reduces auditor

independence. In other words, it is a question of whether the joint provision of NAS is

able to increase or decrease the quality of an audit. 21

Beattie and Fearnley (2002)

claim that there no formal theory of auditor independence and they suggest that most

of the definitions of auditor independence basically highlight the importance of an

auditor‟s objectivity and integrity.

Several studies suggest that the joint provision of audit and NAS may create potential

benefits (see Simunic, 1984; Beck et al., 1988a; Arruñada, 1999a; 1999b; 2000;

Wallman, 1996; Goldman and Barlev, 1974). These studies claim that the joint

provision of NAS enhances an auditor‟s independence through economies of scope

and the auditor‟s economic power. Economies of scope can be defined as the savings

that arise when both services (audit and NAS) are jointly provided by the same

auditors. They can be divided into two types, namely: knowledge spillovers and

contractual economies of scope (Arruñada, 1999a; 1999b). Knowledge spillovers are

derived from the transformation process that occurs when both services need to use a

similar set of information and/or require similar professional qualifications (Arruñada,

1999a; 1999b). For example, when an auditor is undertaking an audit service,

relevant information on the client‟s internal control system and competency in

information technology are both necessary to the performance of the work. A similar

set of information and qualifications are also essential to auditors when they are

providing advice on a client‟s accounting information system. As mentioned by

Wallman (1996), the gathered information flows both ways, when auditors undertake

audit work they learn more about a client's business and this information is also useful

when they provide NAS.

The second type of economies of scope are contractual economies of scope which are

related to the contractual advantages that stem from making better use of assets and/or

Since the definition of audit quality comprises aspects of the auditors‟ competency and their

independence (DeAngelo, 1981; Watts and Zimmerman, 1986), the present study argues that any

factors that are associated with auditor independence will also have an impact on audit quality. NAS

include compliance and assurances services such as taxation, accounting advice, due diligence, internal

audit, training, services to SMEs, corporate recovery and insolvency, legal, forensic and litigation

support, mergers and acquisitions, recruitment and human resources as well as information systems

and technology (Beattie and Fearnley, 2002; Arruñada, 1999a; 1999b).

the safeguarding of an auditor's reputation (e.g. big-size) that has already been

developed through contracting and ensuring quality in auditing or NAS. The use of

similar contractual resources, also referred to as “one-stop shopping”, provides both

supplier and client with advantages that are contractual in nature (Arruñada, 1999a:

775-6). In other words, clients and auditors can reduce the costs of searching and

marketing for those services. Also, clients are assured of the quality of an auditor's

performance. In addition to recognising the cost saving benefits of NAS, Goldman

and Barlev (1974) and Nichols and Price (1976) provide a more complex view of

auditor-client relationships. They suggest that economic power models provide the

auditor with a potential source of power and independence to withstand auditor-client

conflicts. Goldman and Barlev (1974) suggest that the higher the provision of non-

routine NAS, the more important these services are to clients and the greater the

power that an auditor would gain. Such an increase in auditor's power allows them to

be more independent because they are more able to resist the client's pressure.

Despite the positive effects of NAS, their dual provision continues to be controversial

and they are viewed with scepticism due to the potential that they may compromise

auditor independence (Panel on Audit Effectiveness, 2000; SEC, 2001; Beattie and

Fearnley, 2002) and cause economic dependence on clients (Beck et al. 1988a, b;

Simunic, 1984). The foregoing evidence consistently suggests that both the user and

the investor may perceive that NAS impaired audit independence (Wines, 1994; Firth,

2002; Frankel et al. 2002; Raghunandan, 2003; Sharma and Sidhu, 2001; Larcker and

Richardson, 2004) and this outweighs their positive effect of NAS and reflects a

declining in audit quality.

The first argument against the joint provision of NAS is the probability that it will

make auditors economically dependent on their clients. As such, they are less likely to

resist a client's pressure for the fear of losing them. As pointed out by Simunic (1984)

and Beck et al. (1988a), the joint provision of audit and NAS creates knowledge

spillovers or cost savings that lead to economic rents. In order to retain their clients,

auditors tend to accept a client‟s preferences, and consequently they lose their

objectivity. The second argument concerns the nature of NAS themselves and

maintains that they may reduce an auditor‟s independence because of the conflict of

interest that arises during the provision of audits and NAS. For example, they may

audit their own work, take a managerial role or act on behalf of a client‟s management

in an adversarial situation and, therefore, become too close to a client's management

(Panel on Audit Effectiveness, 2000; SEC, 2001; Beattie and Fearnley, 2002).

In sum, generally, prior empirical studies provide inconsistent findings on the

relationship between NAS and auditor‟s independence. Hartley and Ross (1972) claim

that NAS are a minor problem in the total auditor independence issue.22

Some studies

argue that NAS have little impact on auditor independence (Ryan, 2001; Craswell,

1999) and a few suggest that NAS provide feasible advantages (Lai and Krishnan,

2009).23

A number of empirical studies have been unable to find any association

between NAS and auditor independence (Barkess and Simnett, 1994; Craswell, 1999;

Ashbaugh et al., 2003; Chung and Kallapur, 2003; DeFond et al., 2002), whereas

other studies provide evidence that the joint provision of NAS impairs auditor

independence (Wines, 1994; Firth, 2002; Frankel et al. 2002; Raghunandan, 2003;

Sharma and Sidhu, 2001; Larcker and Richadson, 2004).

In this thesis, consistent to the regulatory concern that NAS could compromise the

auditor independence, the higher provision of NAS is indicate as a lower audit quality.

3.3.3 Industry specialist auditors

The theoretical foundation for the use of industry specialist auditors is derived from

the reputation capital hypothesis as it applies to big-size auditors. Economic theories

of product differentiation suggest that sellers expend their resources to build a

reputation (Klein and Leffler, 1981; Shapiro, 1983). In the audit market, there are two

levels of reputation development. The first level requires audit firms to invest in their

brand name's reputation in order to differentiate the quality of their products. The

Hartley and Ross investigate the perceptions of three groups of respondents on NAS, namely:

auditors, financial analysts and management. Their results show that only 5.6% of the respondents rank

the NAS provision as important threat to auditor independent. The overall findings suggest that that

auditor independence in fact can be retained while providing NAS, but it is not essentially improved

the auditor independence in appearance.

23 For example, Lai and Krishnan (2009) findings suggest that when the firms purchase the related

services of financial information system (FIS) from the external auditors, they are more likely to have

higher market value of equity, suggesting that the FIS related services are been priced by investors as a

value-added activity. Such belief shows that NAS increased the firm value, thus contribute to the

efficiency and effectiveness of the overall firms‟ performance, despite raising a threat to auditor

independence.

second level requires big-size auditors to actually differentiate the quality of their

products among them. Due to the complexity and unique features of certain industries,

buyers demand industry specialist auditors in order to deal with specific accounting

rules and reporting requirements (Craswell et al., 1995). Such demands encourage

big-size auditors to invest resources in a particular industry in order to gain industry

specific knowledge and competency.

Evidence suggests that specific knowledge, task-specific experience and training

increase an auditor‟s competence (Ashton, 1991; Boner and Lewis, 1990) and results

in auditors seizing increasing numbers of established audit clients in specific

industries. The auditor market share rises as the number of their audit clients

increases. The larger an auditor's market share is, the more likely clients are to

perceive that that auditor would supply a higher quality audit. This notion is

consistent with the studies that show that a firm‟s market shares signals their product's

quality (Smallwood and Conlisk, 1979; Shockley and Holt, 1983; Caminal and Vives,

1996).

The industry specific knowledge and competency that is possessed by an auditor

represents the main component of their audit quality. Taylor (2000) and Low (2004)

claim that an auditor‟s knowledge of a clients‟ specific industry affects the level of

audit risk assessment and other audit-planning decisions. When auditors have a higher

knowledge and a better understanding of the clients‟ industries they are able to assess

appropriately the levels of audit risk and to plan their audit strategies, and this can

help them to anticipate the potential for misstatements

Evidence has also suggested that the possession of industry specialist knowledge

improves auditor performance. Owhoso et al. (2002) examine the effectiveness of

industry specialist auditors in detecting errors during the audit review process for two

specific industries, namely banking and health care. Their findings suggest that the

auditors‟ experience in the specific industry enables them to detect error more

effectively than non-specialist auditors. Auditors without specific industry experience

perform below the nominal benchmark for error detection. Similarly, Bédard and

Biggs (1991) show that the auditors who have greater manufacturing experience are

better at detecting errors than the auditors who have less manufacturing experience.

Dunn and Mayhew (2004) find a positive relationship between industry specialist

auditors and disclosure quality. Their findings suggest that auditors with specific

industry knowledge are more able to assist their clients in developing industry specific

disclosure strategies. O‟Keefe at al. (1994) find that industry specialist auditors are

associated with a higher compliance with GAAS reporting standards than non-

specialist auditors. Carcello and Nagy (2004) report a negative relationship between

industry specialist auditors and the incidence of fraudulent of financial reporting, and

this indicates that industry specialist auditors are less likely to be associated with

financial fraud.

Several studies have linked auditor industry specialists with earnings management

(e.g. Balsam et al., 2003; Krishnan, 2003a). These studies report that the clients of

industry specialist auditors experience lower discretionary accruals than the clients of

non-specialist auditors. The findings suggest that industry specialist auditors are more

likely to constrain earnings management and the opportunistic behaviour of

management.

In other studies, concerning market reaction, Knechel et al. (2007) claim that when

firms switch from a Big 4 non-specialist to a Big 4 specialist auditor, those firms

experience a significant positive abnormal return. Subsequently, the markets react

negatively when firms switch from a Big 4 specialist to a Big 4 non-specialist. These

findings indicate that the market perceives audit quality differences based on industry

specialization.

Beside the theoretical justification and empirical evidence for the connection between

industry specialist auditors and audit quality, regulators and authoritative guidance

have also emphasised the importance of an auditor being able to understand the

client‟s industry setting before proceeding with auditing work (Knechel et al., 2007).

For example, the UK Auditing Standard, ISA 300: Planning an audit of financial

statement (ABP, 2004), states that an auditor needs to establish an understanding of

the client‟s industry setting before planning their audit strategies.

In summary, most of the prior studies indicate that an auditor's industry knowledge is

a crucial component in the efficiency and effectiveness of audit processes and that it

increases the quality of auditing services. The use of an industry specialist auditor not

only improves the quality of auditing work but is also perceived to be valuable to

market participants.

3.4 The effectiveness of boards of directors

Fama and Jensen (1983) suggest that the board of directors is the highest-level of

control mechanism in an organization because they possess the ultimate power to

compensate the decisions that are made by the top management. Evidence suggests

that several characteristics of a board may influence their effectiveness in their

monitoring role. These characteristics are: size, composition of independent non-

executive directors, financial expertise and meeting frequency.

3.4.1 Board of directors size

Board size is believed to be a fundamental aspect of effective decision making (UK

Corporate Governance Code, 2010). Vafeas (2005) suggests that the size of a

committee and the performance of the directors have a non-linear relationship. Both

too small and too large a size of board is likely to make it ineffective. Lipton and

Lorsch (1992), for example, recommend the ideal size for a board should not exceed

eight or nine directors. Jensen (1993) claims that when a board has beyond seven or

eight members, it is less effective due to the problems of coordination and process

which, in turn, contribute to weak monitoring.

Although average board sizes are relatively large, prior studies have shown that

smaller boards are more effective as directors can communicate better on them and

they are easier to manage.24

These factors promote a more resourceful conversation.

For example, studies of board size and corporate performance have indicated that

smaller boards are associated with higher market values. Yermark (1996) examines

452 large U.S firms in the period between 1984 and 1991 and he documents a

negative relationship between board size and firm value. Drawing from Yermark‟s

Xie et al., (2003) and Vafeas (2005) report the average size of board in their samples are 12.48 and

11.6, respectively.

study, Eisenberg et al. (1998) also provide a similar conclusion on board size and firm

value in a sample of small and mid-size Finnish firms.

In studies related to earning informativeness, Vafeas (2000) claims that market

participants perceived the information content of earnings as being higher in the firms

with a smaller board (with a minimum of five members). This is probably due to the

commitment of each individual member and the likelihood of them accepting their

responsibility as a personal obligation. By comparison, on a larger board, the

responsibility of monitoring is divided among members and less responsibility is

carried by each individual member (Vafeas, 2000).

With regard to audit quality studies, Abbott et al. (2004) suggest that the firms with

smaller board size experience a lower incidence of restatements as the smaller boards

contribute to effective communication and there is less likelihood of a communication

breakdown. This suggests that when board members communicate effectively, they

reduce the incidence of misunderstanding and consequent errors, and that they are

more sensitive to the issues that may affect their shareholders or investors‟

confidence, particularly concerning financial reporting issues.

3.4.2 Independent board of directors

Non-executive directors are associated with the responsibility for monitoring

managers and thereby reducing agency costs that arise from the separation of

ownership and control in day-to-day company management (Fama, 1980; Fama and

Jensen, 1983; Brennan and McDermott, 2004). The UK Corporate Governance Code

(2010: 11) highlights that one of the main responsibilities of non-executive directors

is to “satisfy themselves on the integrity of financial information and that financial

controls and systems of risk management are robust and defensible”. Thus, higher

proportions of independent non-executive directors on boards are expected to induce a

more effective monitoring function which then leads to more reliable financial

statements. This is due to the incentive for independent board members to develop

reputations as experts in decision making (Fama and Jensen, 1983) and to provide an

unbiased assessment of a management‟s actions (Vance, 1983).

Prior studies indicate that an independent board is an effective monitoring safeguard.

Beasley‟s (1996) study suggests that larger proportions of non-executive directors on

boards tend to be negatively related to financial statement fraud. O‟Sullivan (2000)

and Carcello et al. (2002) document a positive relationship between the proportion of

non-executive directors on a board and audit quality. This suggests that independent

board members demand more in-depth audit efforts from the auditor, leading to a

higher quality audit. Similarly, a stream of literature on earnings management and

independent boards suggest that the firms with a higher percentage of independent

board members encounter a lower incidence of earnings management (Klein, 2002;

Davidson et al., 2005; Peasnell et al., 2005; Xie et al., 2002). In summary, all of these

studies recognise the independent board as facilitating effective monitoring.

3.4.4 Board of directors expertise

Board of director knowledge and experience are important elements in ensuring the

effectiveness of a board's monitoring function. Carcello et al. (2002) suggest that the

members of boards of directors who have more experience in terms of a higher

number of directorships are more likely to demand high-quality audit work. Further,

Chtourou et al. (2001) claim that the directors with a higher tenure of board

experience are less likely to be associated with earnings management. Both studies

conclude that higher levels of board expertise lead to a higher monitoring incentive.

In addition, when the board of directors is financially literate they can understand and

address issues relating to financial statements. Xie et al. (2003) find that earnings

management is less likely to occur in firms that are run by a board of directors who

have a corporate and financial background. Similarly, Agrawal and Chadha (2005)

claim that the likelihood of earnings restatement is lower in firms whose boards of

directors are financially literate. Evidence from auditor independence literature

suggests that boards of directors with financial expertise tend to limit the NAS

purchased from auditors as they believe that a provision of NAS compromises auditor

independence (Lee, 2008).

In summary, all of these studies recognise that the boards of directors who have

specific knowledge and experience are useful in monitoring both management and

auditors. Since this thesis examine the audit quality and earnings management, the

accounting and financial knowledge are beneficial to board of director to understand

better the financial statements and issues related to financial reporting.

3.4.4 Board of directors meetings

One of the responsibilities of director is attending meeting and by doing so they

would have privilege to vote on key decisions (Ronen and Yaari, 2008). Conger et al.

(1998) suggest that more frequent board meetings improve a board‟s effectiveness.

The meetings are a key dimension of board operations (Vafeas, 1999) and an indicator

of the effort put in by the directors (Ronen and Yaari, 2008). Active boards that meet

more frequently are more likely to perform their duties in accordance with

shareholders‟ interests (Vafeas, 1999) and to put more effort into monitoring the

integrity of financial reporting and thereby improving the audit quality.

Vafeas (1999) finds that a higher meeting frequency is a reaction to deteriorating

performance. Xie et al. (2003) argue that a board that rarely meets may only have time

for signing management plans and listening to presentations and, therefore, time to

focus on issues such as earnings management will be limited. This shows that board

activity affects performance and it is an important factor in constraining earnings

management.

Carcello et al. (2002) and Krishnan and Visvanathan, (2009) suggest that a higher

frequency of board meetings leads to higher audit fees, and this is consistent with the

argument that when a board of directors meets more frequently, they demand an

extensive audit effort from the auditor, which improves the quality of the audit

process. In addition to this, Chen et al., (2006) examine the 169 firms under Chinese

Securities Regulatory Commission (CSRC) enforcement actions between the period

of 1999 to 2003. They suggest that the higher frequency of board meetings reduces

the likelihood of fraud because regular meetings allow the directors to identify and

resolve potential problems, particularly those that are related to the quality of financial

reporting.

3.5 The effectiveness of audit committees

The audit committee is one of the committees that is established by the board of

directors and whose main responsibility has to do with financial reporting. Apart from

the benefit that is gained from the establishment of an audit committee, prior studies

suggest that the size, composition, expertise and meeting frequency of audit

committees may impact their monitoring effectiveness (e.g. DeZoort et al., 2002;

Walker, 2004).

3.5.1 Audit committee size

According to the UK Corporate Governance Code (2010: 19), “the board should

establish an audit committee of at least three independent non-executive directors.”

(C.3.1). It seems that the size is also one of the important characteristics that

contribute to an audit committee‟s effectiveness.

Consistent with the argument for an effective committee size, too small a committee

size may mean that an insufficient number of directors are able to serve the committee

and thereby the effectiveness of monitoring is decreased (Vafeas, 2005). This is

probably due to an individual director being unable to perform their duties efficiently

as the committee‟s assignments are spread across a small number of directors. In

addition to this, when a committee is too large, the performance of the directors may

decline due to the problems of coordination and process and therefore this is identified

as another cause for weak monitoring (Jensen, 1993; Vafeas, 2005).

The ideal average size of an audit committee is between 3 and 4 members (e.g.

Vafeas, 2005; Xie et al., 2003; Abbott et al., 2004). Evidence from audit committee

sizes suggests that firms with larger audit committees are more effective in

monitoring management. Yang and Krishnan (2005) examine the relationship

between audit committee size and quarterly earnings management in 896 US firms in

the years 1996 to 2000. They find that quarterly earnings management is lower for

firms that have a larger audit committee size.25

This may suggest that having a

sufficient number of audit committee members increases the efficiency of an audit

committee‟s function in terms of monitoring the integrity of financial reporting.

Yang and Krishnan (2005) report the mean, first quartiles, median and third quartile of audit

committee size as 3.587, 3, 3 and 4, respectively.

In another study, Chen and Zhou (2007) find that the firms with larger audit

committees are more concerned about their auditors‟ reputations and tend to assign

the Big 4 auditors. In summary, the larger the size of the audit committee, the more

effective they are in monitoring financial reporting.

3.5.2 Independent audit committee

Agency theory suggest that the independence of a non-executive director is an

essential quality that contributes to a committee‟s effective monitoring function

(Fama and Jensen, 1983) and that the empirical evidence on audit committee

independence is consistent with this proposition. Several studies suggest that

independent audit committees are less likely to be associated with financial statement

fraud (Abbott et al., 2004; Abbott et al., 2000) and more likely to be associated with

lower earnings management (Klein, 2002; Xie et al., 2003; Bédard et al., 2004;

Davidson et al., 2005) and a lower incidence of earnings restatement (Agrawal and

Chadha, 2005) . The independent audit committee is expected to provide unbiased

assessment and judgement and to be able to monitor management effectively.

In addition, Carcello and Neal (2000) provide evidence of the relationship between

audit committee independence and the disclosure choices of firms in financial

distress. They suggest that firms with a higher percentage of independent audit

committees are less likely to receive a going concern audit opinion from auditors.

Further, Carcello and Neal (2003) suggest that independent audit committees are more

effective in protecting auditors from dismissal following the issuance of a going-

concern audit report.

In a study related to auditor quality, Abbott and Parker (2000) and Chen et al. (2005)

suggest that having a higher proportion of independent non-executive directors in

audit committees increases the tendency to assign industry-specialist auditors. In

summary, all of these suggest that independent of audit committees are associated

with a higher quality of financial reporting and that they can be regarded as effective

monitors.

3.5.3 Audit committee expertise

According to the UK Corporate Governance Code (2010: 19), “the board should

satisfy itself that at least one member of the audit committee has recent and relevant

financial experience.” DeZoort (1998: 17) argues that an audit committee member‟s

experience in accounting and auditing is necessary for a sufficient understanding of

oversight tasks. He suggests that:

“…audit and internal control evaluation experience makes a difference in audit

committee members‟ performance on an internal control oversight task. Of

primary importance, audit committee members with experience made internal

control judgements more like those of experts (i.e. practising auditors) in the

area than did members without experience.”

In other words, the regulatory concern and the experimental evidence suggests that

having appropriate experience and knowledge, particularly in accounting and

auditing, is likely to improve an audit committee‟s performance and judgement.

The empirical evidence from archival studies also suggests that the financial expertise

of an audit committee improves its monitoring ability and results in an increase in the

quality of a firms‟ financial reporting. Krishnan and Visvanathan (2008) examine the

association between the financial expertise of audit committees and financial

reporting quality, measured by the level of accounting conservatism, in a sample of

929 US firms from 2000 to 2002. They argue that, with financial expertise in

accounting, audit committees can efficiently assess the nature and the appropriateness

of accounting choices, constrain the aggressiveness of accounting policies and provide

incentives to avoid the risk of litigation. Their findings suggest that audit committees

with accounting financial expertise increased the overall audit committee‟s oversight

function and thus they were more likely to promote accounting conservatism than the

audit committees with non-accounting or non-financial expertise, particularly in an

environment where the board of directors was strong.

Similarly, DeFond et al. (2005) find that market participants react positively to the

appointment of an audit committee with financial expertise in accounting, but no

reaction is noted for audit committees with non-accounting financial expertise.26

is due to the fact that the appointment of committee members with accounting

financial expertise improves the oversight function of the committees and thus

provides a credible signal to the investors that the firms aspire to a higher quality of

financial reporting. In addition, DeFond et al. (2005) suggest that positive market

reaction is concentrated on the firms that are relatively strong in corporate

government. Both studies conclude that an audit committee with financial expertise

complements a strong corporate governance environment by improving the board‟s

ability to protect shareholder interests and increase their firm‟s value.

It has been suggested that an audit committee‟s financial expertise is associated with a

higher quality of financial reporting (Carcello et al., 2002; Abbott et al., 2003a) and

less likelihood of opportunistic earnings (Xie et al., 2003; Bédard et al., 2005). The

reason for this is the financial knowledge and experience that improves an audit

committee‟s oversight function and its ability to facilitate effectively the financial

reporting process. Overall, the empirical evidence reviewed supports the proposition

that the audit committee with financial expertise has improved their effective

monitoring function.

3.5.4 Audit committee meeting

Several studies suggest that firms with a higher number of audit committee meetings

experience less financial restatement (Abbott et al., 2004), are less likely to be

sanctioned for fraud and aggressive accounting (Abbott et al., 2000; Beasley et al.,

2000) and are associated with a lower incidence of earnings management (Xie et al.,

2003). These studies suggest that audit committees who meet regularly during the

financial year are linked to effective monitoring. The more frequently they meet, the

more efficiently they discharge their oversight responsibilities.

DeFond et al. (2005) and Krishnan and Visvanathan (2008) measure the audit committee expertise in

three ways: accounting financial experts (directors with experience as certified public accountant,

controller or chief accounting officer), nonaccounting financial experts (director with experiences as

chief executive officer or president) and nonfinancial experts (directors who are neither accounting nor

nonaccounting financial experts).

In another study, Krishnan and Visvanathan, (2009) found a positive relationship

between audit committee meetings and audit fees, suggesting that the firms with a

higher number of audit committee meetings demand more assurance and a higher

quality audit from their auditors. In order to provide more assurance and a higher

quality audit, the auditors may need to perform additional audit work in terms of

enlarging the scope of the audit and increasing the levels of audit testing, which

results in both higher audit fees and a higher audit quality. Therefore, the higher the

frequency of an audit committee‟s meetings, the more their monitoring function is

improved.

3.6 Earnings management

According to Watt and Zimmerman (1990) and Fields et al. (2001), earnings

management may derive from the flexibility of accounting choices that are given by

GAAP. The GAAP allows managers to choose the appropriate reporting procedures

and to make estimations and assumptions according to their business environment.

Moreover, with an alternative on offer, the manager may choose the reporting

procedure that could benefit and increase the wealth of all the contracting parties

(Watt and Zimmerman, 1990). As a result, accounting choices may create the problem

of earnings management.27

Such a problem, for example, causes shareholders,

investors and debt holders to be unable to distinguish the true economic value of a

firm because their reports do not accurately reflect the actual performance of the firm.

Schipper (1989) defines earnings management “in the sense of purposeful

intervention in the external financial reporting process, with the intent of obtaining

some private gain”. Healy and Wahlen (1999) claim that earnings management occurs

when the managers use their judgement in preparing financial statements with the

intention not to report the firm‟s actual economic performance or in order to gain

benefit from the „adjusted figure‟. Consistent with this definition and description of

earnings the present study views earnings management as the opportunistic behaviour

of management.28

As pointed out by Fields et al. (2001: 260), “not all accounting choices involve earnings

management, and the term earnings management extends beyond accounting choice, the implications

of accounting choice to achieve a goal are consistent with the idea of earnings management”.

28 The present study acknowledges that earnings management can also be regarded as beneficial

information by the market. The managers may use earnings to communicate private information that

Managers engage in opportunistic earnings for several reasons such as for bonus

compensation (Healy, 1985; Gaver at al., 1995; Holthausen et al., 1995), the

avoidance of debt-covenant violation (Sweeney, 1994; DeFond and Jiambalvo, 1994),

the prevention of earnings decreases and losses (Bugstahler and Dichev, 1997; Barth

et al., 1999) and compensating for regulatory or political costs (Cahan, 1992; Jones,

1991; Han and Wang, 1998). Each of these motivations is now reviewed.

Agency theory suggests that one way to monitor an agent‟s behaviour is through their

compensation contracts, enabling the interest between principal and agent to be

perfectly aligned (Jensen and Meckling, 1976). Such contracts, for example, can be

formed between shareholders and managers or as debt-covenants between managers

and lenders. Since compensation contracts and debt-covenants are usually linked to an

accounting number, such contracts create incentives for managers to manipulate

earnings (Watt and Zimmerman, 1978).

Healy (1985) posits that managers tend to choose income-decreasing accruals when

their bonus plans are binding at the maximum or minimum levels and income-

increasing accruals when their bonus plans are not binding. She argues that when

earnings are extremely low and cannot meet the earnings‟ target within the accounting

procedures, the managers have incentives to further reduce current earnings in terms

of differing revenues or accelerating write-offs and these approaches are referred to as

“taking a bath” (Healy, 1985: 86). These actions, however, do not necessarily affect

current bonus awards but may help to meet a future earnings target. In contrast, Gaver

et al. (1995) find the managers select income-increasing accruals when the bonus

plans are falling below the lower bound, and vice-versa. They claim that their results

are more consistent with the income smoothing hypothesis that states that managers

manipulate earnings in order to reduce the divergence of reported earnings and to

could potentially maximise a firm‟s value (Watt and Zimmerman, 1986; Holthausen, 1990;

Subramanyam, 1996; Sankar and Subramanyam, 2001; Arya et al. 2003; Louis and Robinson, 2005).

However, since earnings management is involved a higher degree of managerial judgment, the auditors

and boards of directors safeguard themselves by constraining earnings management activities. Thus, the

present study concentrates on the negative aspect of earnings management.

ensure that the current earnings reach the normal or expected earnings target.29

Holthausen et al. (1995) find similar findings to Healy (1985), but find no evidence

that income-decreasing accruals are associated with a lower minimum boundary and

claim that Healy‟s results may be sensitive to the particular model used to estimate

discretionary accruals. In summary, these studies conclude that bonus scheme plans

create incentives for managers to manipulate earnings in order to maximise their

bonus award.

Sweeney (1994) examines a sample of firms prior the violation of accounting-based

restrictions in debt agreements. She finds that when the managers of defaulting firms

are approaching covenant violation they are more likely to report income-increasing

accruals in order to offset their debt constraints. DeFond and Jiambalvo (1994)

provide similar findings and they conclude that debt-covenants influence a manager‟s

accounting decisions in the preceding year and during the year of the violation.

In general, market participants and stakeholders appear to reward the firms with

positive or higher earnings more than the firms with negative or lower earnings, and

therefore managers manipulate earnings to meet these expectations. Bugstahler and

Dichev (1997) provide evidence that managers manipulate earnings in order to avoid

earnings decreases and losses. Specifically, their results on the frequency of

distributions in earnings show that there are low frequencies of small earnings

decreases and small losses but that higher frequencies of small increases in earnings

and small positive incomes are unusual. Barth et al. (1999) suggest that the firms with

patterns of increasing earnings are more likely to have larger earnings multiples (e.g.

higher coefficient of earnings).

In studies related to regulatory or political cost, Cahan (1992) finds that the managers

of firms under anti-trust investigation report income-decreasing accruals during

Mosses (1987: 360) define income smoothing as “as an effort to reduce fluctuations in reported

earnings. Researchers generally have agreed that smoothing involves the use of some device to reduce

the divergence of reported earnings from an earnings number that is “normal” or expected” for the

firm.”

investigation years. Similarly, Jones (1991) suggests that managers tend to report

income-decreasing accruals during the year of an application for import relief in order

to obtain import relief benefits such as increases in tariff or reductions in quota. Han

and Wang (1998) examine opportunistic earnings in two groups of firms: petroleum

refining firms and the crude oil and natural gas firms during the 1990 Persian Gulf

crisis. Their findings suggest that petroleum refining firms used income-decreasing

accruals to mitigate the possibility of adverse political actions.

In summary, the motivations reviewed above may reflect opportunistic behaviour on

the part of management. This evidence suggests that managers use their discretion to

manipulate reported earnings, and thus the monitoring roles of the board of directors,

the audit committee and the external auditors are required in order to constrain

earnings manipulation behaviour.

3.7 Accruals-based measure of earnings management

There are several potential instruments can be employed by managers to manipulate

earnings management. This includes the flexibility of accounting method, income

smoothing and accrual accounting. Among others, the managements are more favour

toward the accruals accounting due to low cost and difficult to observe (Young,

1999).

The accruals can be divided into two components: the discretionary accruals and non-

discretionary accruals. The discretionary accruals also known as abnormal accrual or

managed accruals, which always related to earnings manipulation. While, the non-

discretionary accruals referred as normal accruals or non-managed accruals. These

terms are used interchangeably in earnings management studies (Kang and

Sivaramakrishnan, 1995).

McNichols (2000) identify three main measures of discretionary accruals in the prior

literature. These include the aggregate accruals models, specific accruals models and

the frequency distribution approach. Several models are introduced in relation to the

aggregate accruals such as Healy‟s (1985) model, DeAngelo‟s (1986) model, Jones‟s

(1991) model, the modified Jones model from Dechow et al. (1995) and the

performance-adjusted discretionary accruals model by Kothari et al. (2005). The main

differences between the models are how the researcher partitions the non-

discretionary accruals component from the total accruals and their ability to

accommodate changes in firm‟s economic condition. The Healy (1985) model and the

DeAngelo (1986) model assumed that non-discretionary accruals are constant, and

these restrictions are seen to be unrealistic because accounting accruals change in

response to economic conditions (Kaplan, 1985). As an alternative, Jones‟ (1991)

model, the modified Jones model by Dechow et al. (1995) and the performance-

adjusted discretionary accruals by Kothari et al. (2005) controls the variations of non-

discretionary accruals by taking into account the changes in total assets, revenues,

receivables as well as the firm‟s performance (e.g. return on assets). In fact, the Jones

(1991) and modified Jones models are recognised in the literature as the most

powerful models for detecting earnings management (Dechow et al., 1995; Young,

1999).30

However, the limitation of aggregate models is the risk of misspecification

when they inefficiently isolate the discretionary component of total accruals.

In relation to specific accruals, the discretionary accruals are an estimate based on

single accruals. Examples of specific accruals models include the residual provision

for bad debt (McNichols and Wilson, 1988), the loss reserves of property and casualty

insurers (Petroni, 1992), loan loss provisions (Wahlen, 1994; Collins et al., 1995;

Beaver and Angel, 1996) and tax expenses (Philips et al., 2003). McNichols and

Wilson (1988) claim that when specific accruals represent a small part of the

discretionary component, they may fail to reflect earnings management in cases where

other discretionary components are manipulated. Thus, stated differently, the

aggregate accruals models give rise to more comprehensive research design in

capturing the discretionary components.

The frequency distribution approach focuses on the behaviour of earnings where there

is a specific intention (e.g. to avoid earnings decreases or losses) or certain thresholds

(e.g. to report positive profits, sustain recent performance, and meet analysts‟

forecast). This approach was developed by Burgstahler and Dichev (1997) and

Dechow et al. (995) evaluates the ability of the several alternative models such as Healy (1985),

DeAngelo (1986), Jones (1991), modified Jones and the industry model. While, Young (1999)

evaluates the Healy (1985), DeAngelo (1986), modified DeAngelo, Jones (1991) and modified Jones

models.

Degeorge et al. (1999), respectively. McNichols (2002) claims that the distribution

approach provides specific predictions related to which firms will manage earnings

rather than merely measuring the magnitude of managers‟ opportunistic earnings. In

other words, the frequency distribution approach cannot infer earnings management

activities, which are the main concern of this thesis.

In summary, the designs of earnings management research are various and the

advantage of each approach is relatively dependent on the research question.

According to McNichols (2000: 333), if the aim of research is to examine the

magnitude of earnings management, the aggregate approach is more appropriate

because specific accruals are relevant to tests for the associations between specific

accruals and hypothesised factors and this requires a researcher to model each accrual

component according to the hypothesised factors. In addition, the findings of specific

accruals are difficult to generalize when specific accruals are not sufficiently

sensitive. On the other hand, the frequency distribution approach cannot be used to

identify the magnitude or the level of opportunistic earnings. The aim of this thesis is

to examine the relationship of effective boards, audit committees and auditor quality

in constraining earnings management. Thus, the magnitude of earnings or the activity

levels of management towards opportunistic earnings are a crucial component of the

investigation. Therefore, for the purposes of this thesis, the aggregate approach is

more appropriate when compared with the specific accruals approach and the

frequency distribution approach. There are three models of aggregate accruals that

will be employed: the cross-sectional Jones (1991) model, the cross-sectional

modified Jones model and the performance adjusted discretionary accruals model.

These models will be explained in Chapter four.

demand for audit quality

The following section reviews prior studies that have examined the relationship of

effective boards and audit committees to the various proxies of audit quality (e.g.

engagement of Big-size auditors, industry specialist auditors, restatements, fraud,

litigation against auditors, auditor tenure, the appropriateness of going concern audit

opinion, audit fees and NAS fees). These studies suggest that several characteristics of

boards and audit committees, reviewed in section 3.4 and 3.5, are linked with

effective monitoring that enhances the overall quality of financial reporting, and

particularly the quality of audit services.

Several studies suggest that boards of directors and audit committees may influence

the selection of an external auditor (Knapp, 1991; Beasley and Petroni, 2000; Chen et

al., 2005; Abbott and Parker, 2000; 2002). The selection criteria that are used are

based on the skills and abilities of auditors in enhancing the audit process. Knapp

(1991) examines the behaviour of audit committee members and their choice of

external auditors. He claims that audit committees appear more likely to choose Big 8

auditors than non-Big 8 auditors because the Big 8 auditors are inclined to report any

material misstatements that they discover during their auditing work. Moreover, he

suggests that audit committee members perceive that during the early years of an audit

engagement there is a gradual improvement in audit quality due to a „learning curve

effect‟. However, audit committee members also tend to perceive that during the

subsequent years of an auditor-client relationship the audit quality may progressively

reduce because that relationship can impair an auditors‟ independence.

Abbott and Parker (2000), Beasley and Petroni (2001) and Chen et al. (2005) examine

more specific characteristics of boards and audit committees with respect to the

selection of industry specialist auditors. As far as the present study is concerned, these

are the only studies in this area that are based on US and Australian samples. There is

no study that examined UK firms. Industry specialist auditors are more desirable

because they are more reliable than non-specialist auditors for detecting errors

(Bédard and Biggs, 1991; Wright and Aright, 1997) and frauds (Johnson et al., 1991;

Carcello and Nagy, 2004). Abbott and Parker (2000) investigate the proportion of

independent non-executive directors on boards and audit committees as well as the

audit committee meetings. They suggest that the audit committees with solely

independent non-executive directors that meet at least twice a year are more likely to

employ industry specialist auditors. They also report insignificant relationship exists

between the proportion of independent non-executive directors on boards and the

employment of industry specialist auditors. Using a more specific sample, Beasley

and Petroni (2001) claim that the property-liability insurers that have a higher

percentage of non-executive directors on their boards tend to employ industry

specialist auditors.31

Chen et al. (2005) investigate the characteristics of the boards

and audit committees of the top 500 Australian firms. They suggest that an audit

committee with a higher percentage of non-executive directors is more likely to

employ industry specialist auditors. However, they find no significant relationship

between the audit committee meetings and expertise and the employment of industry

specialist auditors.

Abbott et al. (2004) claim that financial restatement may signal inefficiency of

financial reporting because it indicates that auditors have failed to identify errors in

prior financial statement. Such inefficiency can be regarded as being indicative of a

lower quality of both financial reporting and auditing (Kinney et al., 2004). Abbott et

al. (2004) suggest that audit committees with independent members that are active and

have financial expertise are more efficient in monitoring the financial reporting

process, and that this leads to fewer occurrences of financial restatement. Consistent

with this evidence, Agrawal and Chadha (2005) find that boards or audit committees

with independent directors who have financial expertise are also associated with a

lower incidence of restated earnings.

With regard to fraud, Beasley (1996) suggests that smaller boards‟ size and higher

proportion of non-executive directors on boards improve the boards‟ function in

monitoring the top managements‟ behaviour, particularly in preventing the financial

statement frauds.32

However, her finding on the establishment of audit committee is

not significantly related with the incident of fraud, contradicts to McMullen‟s finding.

McMullen (1996) suggest that the establishment of audit committee encourages

higher quality of financial reporting in term of fewer lawsuits for fraud, less quarterly

The inconsistency of this finding with Abbott and Parker (2000) might be due to the different

research design and variable definitions employed in the Beasley and Petroni (2001) study. Abbott et

al. (2000) define industry specialist auditors based the three specific measurements that are employed

in the Palmrose (1986), Craswell et al. (1995) and Dopuch and Simunic (1982) studies. These

measurements are based on the audit clients‟ industry sales. Beasley and Petroni (2001) assigned

indicators for dependent variables; value 2 if the auditors are both specialists and Big 6 auditors, value

1 if the auditors are non-specialist Big 6 auditors, and value 0 if the auditor is a non-Big 6 auditor.

32 In her analysis, she also differentiates between grey directors (the non-executive directors who are

related to management) and independent non-executive directors. Her results suggest that both

variables (percentage of grey directors and percentage of non-executive directors) are significant and

have a negative relationship with the occurrence of financial statement fraud. This evidence shows that

the term „independent‟ is insensitive to the definition of non-executive directors. The result for board

size is presented in the supplemental analysis (Beasley, 1996).

earnings restatement, less SEC enforcement and less illegal action. In a similar area of

study under the SEC enforcement samples, Dechow et al. (1996) claim that where the

board of directors is dominated by the management, practising the dual role functions,

CEO is also the founder of the firms, fewer representatives of outside block holders

and no audit committee are more likely to engage in earnings manipulation. Using

more audit committee variables, Abbott et al. (2000) suggest that audit committees

which are comprised of solely independent non-executive directors and which meet at

least twice a year encounter fewer fraudulent financial statements. This finding

supports the effective role of independent non-executive directors as a key to the

monitoring of the financial reporting and auditing process. Chen et al. (2006) examine

the relationship between the board of directors‟ characteristics and financial fraud in

China. They find that the firms with boards of directors that are comprised of a higher

proportion of independent non-executive directors and that have a higher frequency of

meetings are less likely to commit fraud. However their result for board size is

insignificantly related with the incidence of fraud.

Carcello and Neal (2000) examine the relationship between the composition of audit

committees and the likelihood of going-concern audit opinion in 223 firms that

experienced financial distress during 1994. They suggest that the higher the number of

affiliated directors (i.e. grey directors) in the audit committee, the lower the likelihood

of auditors in issuing going-concern audit reports. These imply that the predomination

of affiliated directors in the audit committees are able to influence auditors‟ decision

to issue the audit opinion (i.e. instead of issuing a modified report, the auditor issues

unmodified report) and to dismiss the auditors if they refuse to issue clean reports.

They further suggest that the audit committees with greater independence, equipped

with financial expertise and lower stockholding are more effective in sustaining the

auditor against dismissal after the issuance of new going-concern audit reports

(Carcello and Neal, 2003).

Prior studies on the relationship of boards of directors and audit committees to audit

fees can be viewed from different perspectives. The demand-based perspective

suggests that an effective board of directors and audit committee demands a higher

quality audit and greater assurance from the external auditors in order to protect their

own interests (Carcello et al., 2000). Specifically, Carcello et al. (2002: 369) claim

that a board of directors may seek to purchase a differentially higher audit quality to

obtain an enhanced assurance in order to protect their “reputation capital, avoid legal

liability and promote the shareholder interests”. From this perspective, a higher audit

quality is indicated by higher audit fees, which are consistent with the extensive audit

effort and audit time that is set by auditors while performing their services.

Consistent with agency theory in respect to the vigilant oversight function of non-

executive directors, O‟Sullivan (2000) claims that the firms with a higher percentage

of non-executive directors on boards are more likely to incur higher audit fees.33

similar vein, Carcello et al. (2002) examine the relationship between a board‟s

composition, meetings and directorship, and the level of audit fees. Their findings

suggest that firms with a higher percentage of independent non-executive directors34

more frequent board meetings and higher number of multiple directorships tend to

demand a higher quality audit and a higher level of assurance. They inferred the

demand for different levels of audit quality through the choice of big-size auditors

because the size of auditors indicates different levels of quality (DeAngelo, 1981). A

higher level of assurance is measured by the audit effort; “additional audit works” that

are “beyond the auditors‟ cost-minimizing level” may result in a higher level of

assurance (Carcello et al., 2002: 369). In their additional analysis, Carcello et al.

(2002) replace the board characteristics with audit committee characteristics (i.e.

composition, meetings and expertise). Their results show that audit committee

independence and audit committee expertise are positively related with audit fees.

However, the result for audit committee meetings is insignificantly related with audit

fees. They examine further by integrating both the characteristics of board and audit

committee and find that the results for the board of directors remain unchanged but

that none of the audit committee‟s characteristics are significantly related to audit

fees. This may suggest that, when a board is present, the audit committee‟s function

might reduce as there is an increase in monitoring by the board. One of the limitations

He does not differentiate between the types of non-executive director.

34 Carcello et al. (2002) differentiate the types of directors, i.e. the independent non-executive directors

(non-management directors who are free from any interest) and grey directors (non-management

directors who have an economic or personal interest in the firm). Their results suggest a significant

positive relationship between audit fees and both types of directors (% of grey directors on board and

% of independent non-executive directors on board at p<0.01 and p<0.10 respectively).

identified in Carcello et al. is that they did not consider the problem of endogeneity in

their analysis.

In contradiction to the evidence that is reported by Carcello et al. (2002), Abbott et al.

(2003a) claim that the presence of a board of directors and an audit committee that is

comprised of solely independent non-executive directors and at least one member who

is equipped with financial expertise is positively associated with audit fees.35

also report that the number of independent non-executive directors on board and the

number of board meetings are positively related with audit fees. In another study,

Krishnan and Visvanathan (2009) examine the characteristics of boards and audit

committees and their relationship to audit fees for 801 firm-year observations in the

S&P 500 between 2000 and 2002. They find that the firms with a larger board size

and more frequent board meetings are associated with higher audit fees. They also

suggest a positive relationship between audit committee meetings and audit fees.

As an alternative to a demand-based perspective, a supply-based perspective is based

on the auditor‟s point of view. If auditors perceive that their client is surrounded by

strong corporate governance, this may signal that the firm has effective internal

control and this may reduce the auditor‟s risk assessment and decrease the audit fees.

In order to give an understanding of how an auditor assesses an overall audit risk, the

present study will first explain the audit risk model. SAS 300: Accounting and

Internal Control Systems and Audit Risk Assessments (AICPA, 1995), defines audit

risk as “the risk that auditors may give an inappropriate audit opinion on financial

statements”. Similarly, the International Standard of Auditing, ISA (UK and Ireland)

200: Overall Objective of the Independent Auditor and the Conduct of an Audit in

35 Abbott et al. (2003) highlight that there are several possible explanations for why their results

contradict the findings of Carcello et al. (2002). Firstly, Carcello et al. (2002) employed a survey

method, which may have been exposed to the non-response bias. Secondly, with regard to the type of

sample, Carcello et al. (2002) used Fortune 1000 companies which are actually greater in number than

the total population of SEC registrants. The size of firms used may account for the different findings.

Lastly, the gap between the years of the samples (i.e. Carcello et. al used sample between April 1992

and March 1993, but Abbott used sample between February 2001 and June 2001) may strengthen the

influence of the audit committee‟s role due to increased enforcement by regulators during the period

between the studies, and this may result in dissimilar findings.

Accordance International Standards on Auditing (UK and Ireland), (ABP, 2009: 6)

describe audit risk as “the risk that the auditor expresses an inappropriate audit

opinion when the financial statements are materially misstated.” An audit risk model

contains the elements of inherent risk, control risk and detection risk. Inherent risk is

the risk that is associated with error and misstatement that occurs at the entity, account

balance and class of transaction level. Control risk is the risk that the accounting and

internal control system of an entity is unable to prevent or detect errors in a timely

manner. Detection risk is the risk that an auditor‟s substantive procedures fail to

detect errors and misstatements. Both the inherent risk and the control risk will

determine the detection risk. If an auditor assesses an inherent risk and control risk to

be low then the level of detection risk may be higher, leading to lower level of

substantive procedures. In summary, the assessment of inherent and control risks are

central in determining overall audit procedures.

In respect to a client‟s internal control system, auditors assess the control procedure

and the control environment. The control environment is determined by the overall

“attitude, awareness and actions” of a board of directors and management regarding

internal control and its importance to their organization, while control procedures are

related to policies and procedures that have been established (AICPA, 1995). By

holding the inherent risk constant and applying strong control procedures, it is

possible for a positive control environment (strong board of directors and audit

committee) to reduce both the control risk and the audit risk. Cohen et al., (2002: 579)

point out:

... in the case where a client‟s governance structure has effectively

implemented a strong monitoring as well as strong strategic perspective,

there is the potential for both a more efficient (e.g. less extent of tests of

details) and a more effective (greater assurance of the integrity of the

financial statements) audit.

This may suggest that strong corporate governance promotes an effective internal

control environment. An effective internal control, then leads to a less substantive test

by an external audit and results in lower audit fees.

Several studies have indicated negative relationship between the characteristics of

boards and audit committee with audit fees (for example Tsui et al., 2001; Boo and

Sharma, 2008; Krishnan and Visvanathan, 2009). Tsui et al. (2001) examine CEO

duality roles and audit fees using 650 observations from Hong Kong firms. Their

findings suggest that the firms that separate the roles of CEO and chairman tend to

have lower audit fees, indicating that effective monitoring mechanisms are in place

and that they have the effect of reducing control risk and audit effort. Drawing on the

Tsui et al. (2001) framework, Krishnan and Visvanathan (2009) also suggest similar

findings on CEO duality roles and audit committee expertise. As well as suggesting

that a more effective monitoring role is served by separately functioning CEOs and

chairmen, they also claim that auditors value an audit committee‟s accounting and

financial expertise. The financial expertise of an audit committee reduces the firm‟s

control risk, which in turn is reflected in less audit testing and lower audit fees. Their

findings on audit committee expertise contradict the study done by Abbott et al.

(2003a) that suggests a positive relationship between audit committee expertise and

audit fees. They claim that Abbott et al. (2003a) use a broad definition of audit

committee expertise that includes both accounting and non-accounting financial

expertise. They define accounting financial expertise as the directors having

experience as certified public accountants, auditors, chief financial officers or

financial controllers. The direction of this variable was sensitive to such differences in

the definitions of financial expertise.

Prior discussions assumed that demand-based and supply-based perspectives are

mutually exclusive. However, there is also a possibility that both perspectives could

coexist and that they are not mutually exclusive (Krishnan and Visvanathan, 2009).

For example, when an effective audit committee demands to have a higher quality

audit, it increases both the audit scope and the audit effort. Simultaneously, an

increased effectiveness may also correspond to a strong internal control which is

reflected in the auditor‟s assessment of audit risk (Krishnan and Visvanathan, 2009).

Boo and Sharma (2008) claim that, from the demand-based perspective, the

association between corporate governance and audit fees in regulated companies (i.e.

financial and utilities companies) is weaker because industry specific regulators share

their monitoring and overseeing roles with the external auditor. Thus, they demand

less extensive audit work in the presence of regulatory oversight. From a supply-based

perspective, auditors perceive that when regulators provide an additional oversight

role their presence may decrease the audit risk, thus there is less need for audit testing,

resulting in lower audit fees. They also find that having multiple directorships on

boards or audit committees relates positively to audit fees by encouraging more audit

effort, in order to protect reputational capital, and tends to result in higher audit fees

(in the presence of regulators). Moreover, they claim that auditors perceive audit risk

as being higher due to the time constraints of directors who serve on multiple boards

and this also promotes the need for extra audit work. Goddard and Masters (2000)

investigate two sets of UK data from 1994 to 1995. Their results show that, in 1994,

firms with audit committees have higher audit fees, but the data from 1995 reveals

that there is no significant difference in the level of audit fees between companies,

with or without audit committees. This contradictory result may be due to the

improvements in accounting systems and internal controls that were introduced by the

regulators36

. Similarly, O‟Sullivan (1999) also finds that there is no evidence that a

board of directors‟ and audit committee‟s attributes influence the level of audit fees.37

He explains that such insignificant findings may be due to the effect of monitoring

functions that offset increased audit efforts.

With regard to NAS fees, there are a very limited number of studies that have

examined the relationship between NAS and board of director or audit committee

effectiveness. To date there are only four studies that have investigated these issues,

namely: Abbott et al. (2003b), Lee and Mande (2005), Lee (2008), Adelopo (2010)

and Zaman et al. (2011). All of these are largely based on US firms. Abbott et al.

(2003b) examine the 538 firms that filed with SEC between February 5, 2001 and

March 15, 2001. Using the ratio of NAS fees to total audit fees, they suggest that the

firms with audit committees that are solely independent and that meet at least four

times a year are likely to limit the amount of NAS that are purchased as, in their view,

the higher levels of NAS might potentially impair audit quality. In their further

analysis, they find that audit committee expertise is insignificant with NAS fees ratio.

Lee and Mande (2005) extend the Abbott et al. (2003b) study by modelling the audit

and NAS functions simultaneously. They suggest that the firms with solely

independent committee members who meet at least four times a year have lower rate

The Cadbury Report (1992) applies to the annual reports of the years ending on or after 30 June 1993

for UK public listed companies.

37 O‟Sullivan (1999) uses the 1995 data. According to Goddard and Masters (2000), the year 1995 is a

transition year, as the Cadbury Report (1992) started to take effect from the middle of 1994.

of NAS purchase. However, when they model the NAS fees simultaneously, none of

the audit committee characteristics are significant. Lee (2008) jointly investigates

board of director and audit committee characteristics along with changes in NAS fees

ratios (changes in total NAS fees to total audit fees). He claims that an effective audit

committee (consisting of solely independent members of whom at least one-third have

financial expertise) and board of directors (at least half of whom are independent and

more than the sample median of whom are financial experts) are likely to limit NAS

purchased in order to enhance auditors‟ independence. However, these three studies

do not consider the characteristics of the size of committees or the financial expertise

of board members. Adelopo (2010) examines more a more comprehensive range of

board and audit committee characteristics (except for the financial expertise of boards

of directors) using a simultaneous equation of audit and NAS fees from the FTSE 350

in the periods of 2005 and 2006. He finds that the frequency of audit committee

meetings and the levels of independence on boards are positively related to both audit

fees and NAS fees. In addition, the results suggest the firms with larger board sizes

are likely to have higher NAS fees but pay lower audit fees. Recently, Zaman et al.

(2011) examine the association between the corporate governance quality, audit fees

and NAS fees. The audit fees and NAs fees are measured by natural log of audit fees

and NAS fees, respectively. They find that the larger firms with the effective audit

committee are likely to purchase more NAS due to the complexity of their operations.

Their study, however, do not control the size of board and financial literacy of the

board‟ members.

Overall, the studies of the effects of various characteristics of boards and audit

committees do show that they have a significant impact on audit quality. Most of

these studies, however, are undertaken by US based researchers and the results could

not be generalized due to there being different institutional settings, legal

environments and auditor incentives in other countries. By utilising three

measurements of audit quality (audit fees, NAS fees and the engagement of industry

specialist auditors), part of this thesis examines the relationship of the characteristics

of boards and audit committees (e.g. size, composition, financial expertise and

meeting frequency) to audit quality. As far as the present study is concerned, there is

no prior study that examines the relationship between corporate governance and

industry specialist auditors in the UK. The prior studies that relate audit fees and NAS

fees to corporate governance levels in UK are limited to the examination of several

characteristics of board and audit committee (Collier and Gregory, 1996; O‟Sullivan,

1999, 2000; O‟Sullivan and Diacon, 2002; Adelopo, 2010; Zaman et al., 2011). None

of these studies examine the financial expertise of board members that has been

suggested by the US studies to improve the quality of financial reporting (Xie et al.,

2003; Agrawal and Chadha, 2005). These potential gaps demand further investigation

since UK firms are unique in terms of their voluntary governance system.

Consistent with the evidence and the theoretical bases for the measurement of audit

quality that are provided under section 3.3.1 to 3.3.3, the present study views higher

audit fees (Abbott et al., 2003a; Carcello et al., 2002; O‟Sullivan, 2000)38

, lower NAS

fees (Wines, 1994; Firth, 2002; Frankel et al. 2002; Raghunandan, 2003; Sharma and

Sidhu, 2001; Larcker and Richardson, 2004), and the engagement of industry

specialist auditors (Owhoso et al., 2002; Bédard and Biggs, 1991; O‟Keefe at al.,

1994; Carcello and Nagy, 2004) to be associated with a higher quality audit.

Based on the propositions of agency theory, concerning monitoring roles, and the

evidence from prior literature, the present study posits that a board of directors with a

smaller size, more independence, more financial expertise and more regular meetings

is defined as an effective board. Similarly, an audit committee with more members,

that is solely independent, that possess financial expertise and that meets frequently is

also described as being effective. It is argued that the boards of directors and audit

committees with these characteristics demand a higher quality audit in order to

safeguard their capital reputation, to avoid legal exposure and to promote

shareholders‟ interests. Below are the summaries of hypotheses stated in a form that

uses audit fees, NAS fees and the engagement of industry specialist auditors as

proxies for audit quality:

Despite the proposition that higher audit fees are associated with a higher audit quality and are

consistent with a more extensive audit effort and audit time, the present study is aware of the possibility

that lower audit fees could also associated with a higher audit quality. A supply-based perspective

suggests that if auditors perceive that their clients have strong corporate governance, this may signal

that the firm is supported by effective internal controls which, in turn, may reduce the auditors‟ risk

assessment and decrease the audit fees (Tsui et al. 2001; Boo and Sharma, 2008b; Krishnan and

Visvanathan, 2009).

H1: There is a negative relationship between the board’s size and audit fees.

H2: There is a positive relationship between the board’s size and NAS fees.

H3: There is a negative relationship between the board’s size and the engagement

of industry specialist auditor

H4: There is a positive relationship between the independent board and audit fees.

H5: There is a negative relationship between the independent board and NAS fees.

H6: There is a positive relationship between the independent board and the

engagement of industry specialist auditor

H7: There is a positive relationship between the board’s financial expertise and

audit fees.

H8: There is a negative relationship between the board’s financial expertise and

NAS fees.

H9: There is a positive relationship between the board’s financial expertise and

the engagement of industry specialist auditor

H10: There is a positive relationship between the board’s meeting frequency and

audit fees.

H11: There is a negative relationship between the board’s meeting frequency and

NAS fees.

H12: There is a positive relationship between the board’s meeting frequency and the

H13: There is a positive relationship between the audit committee’s size and audit

H14: There is a negative relationship between the audit committee’s size and NAS

H15: There is a positive relationship between the audit committee’s size and the

H16: There is a positive relationship between the solely independent audit

committee and audit fees.

H17: There is a negative relationship between the solely independent audit

committee and NAS fees.

H18: There is a positive relationship between the solely independent audit

committee and the engagement of industry specialist auditor

H19: There is a positive relationship between the audit committee’s financial

expertise and audit fees.

H20: There is a negative relationship between the audit committee’s financial

expertise and NAS fees.

H21: There is a positive relationship between the audit committee’s financial

expertise and the engagement of industry specialist auditor

H22: There is a positive relationship between the audit committee’s meeting

frequency and audit fees.

H23: There is a negative relationship between the audit committee’s meeting

frequency and NAS fees.

H24: There is a positive relationship between the audit committee’s meeting

frequency and the engagement of industry specialist auditor

3.9 The relationship of board of directors, audit committee and auditor

quality in constraining earnings management

An abundance of studies has examined the monitoring roles of board, audit committee

and auditor quality and their effectiveness in constraining opportunistic earnings.

These studies suggest that an effective board and audit committee and a higher quality

auditor extend their monitoring functions to limit earnings management behaviour.

The following are some of the key papers in this area.

As far as the present study is concerned, there are only six relevant studies that have

been done in the UK. These studies are Peasnell et al. (2000; 2005), Ferguson et al.

(2004), Antle et al. (2006), Kwon et al. (2007), Habbash et al. (2010), Sun et al.

(2010) and Habbash (2010). Peasnell et al. (2000) examine the association between

the size of board and the proportion of non-executive directors on a board with the

incidence of earnings management in UK firms in pre- and post-Cadbury periods.

They find no significant relationship between the number of non-executive directors

on board and earnings management in pre-Cadbury periods, but the results for post-

Cadbury periods suggest that there are fewer incidences of income-increasing accruals

in order to avoid earnings losses or earnings decline when the board of a firm is

comprised of a higher proportion of non-executive directors. The board size is

insignificantly related with earnings management in pre- and post-Cadbury periods.

Peasnell et al. (2005) investigate the effect of the proportion of non-executive

directors on boards, CEO duality and the establishment of audit committees on the

likelihood of earnings management occurring. Their tests are conducted using UK

data from the periods of 1993 to 1996 and they use discretionary accruals as a proxy

for earnings management. They find that the firms with higher percentages of non-

executive directors on board are associated with a lower incidence of income

increasing discretionary accruals, particularly when pre-managed earnings fall below

zero or are less than prior reported earnings. However, there is no evidence to suggest

that CEO duality, board size or the presence of an audit committee have any effect on

the incidence of earnings manipulation. Both Peasnell et al.‟s studies do not consider

endogeneity issues in their models.

Ferguson et al. (2004) provide evidence from UK data for the periods of 1996 to

1998. They examine the likelihood of firms being associated with earnings

management activities (the firms that are criticized by financial analysts or investors

or investigated by the Financial Reporting Review Panel due to alleged accounting

irregularities and opportunistic accounting treatment, and the firms that restate their

prior financial statements or make adjustments under FRS No. 12) and the absolute

value of their discretionary accruals (using the modified Jones model) and their

relationship to NAS. NAS are measured by the ratio of NAS fees to total fees, the

natural log of NAS fees and the percentile rank of NAS fees by practice office. They

find that NAS fees are positively related to earnings management.39

This suggests that

the increased economic bonding between client-auditor may make auditors less likely

to constrain a management‟s opportunistic earnings behavior, and thus reduce the

quality of financial reporting. In addition, they also find that none of the corporate

governance characteristics (i.e. CEO duality roles and the percentage of non-executive

directors on boards) are significantly related to earnings management.

Kwon et al. (2007) provide evidence from an international setting using data from 28

countries, including the UK, from 1993 to 2003.40

Specifically, they examine how the

country‟s legal system affects industry specialist auditors in respect to constraining

Except for when they are measured by the likelihood of firms being subject to irregularities and

opportunistic accounting treatments and by the percentile rank of NAS fees by practice office

40 The 28 countries include Australia, Austria, Belgium, Brazil, Canada, Chile, Denmark, Finland,

France, German, Hong Kong, Indonesia, Ireland, Italy, Korea, Malaysia, Mexico, The Netherlands,

Norway, Philippines, Singapore, South Africa, Spain, Sweden, Switzerland, Taiwan, Thailand and the

United Kingdom.

earnings management. They suggest that the use of industry specialist auditors is

negatively related to discretionary accruals and positively related to the earnings

response coefficient, and that the benefit of employing industry specialist auditors

increases as a country‟s legal system becomes stronger.

Antle et al. (2006) provide evidence from US and UK data. They investigate the audit

fees and NAS fees data from the UK from 1990 to 2004, while from the US data is

from the year 2000. They find that higher NAS fees decreased discretionary accruals,

suggesting that there is a knowledge spillover effect resulting from the joint provision

of audit and NAS. Specifically, they employed the natural log of NAS fees as the

measurement of auditor independence and the Jones (1991) model to detect earnings

management. In relation to audit fees and earnings management, they find a positive

relationship between audit fees and earnings management. They argue that higher

audit fees may create a bias in the client-auditor relationship, thus inclining the

auditor to allow opportunistic earnings behaviour. However, this study did not

incorporate board and audit committee characteristics, which are also believed to

influence the quality of reported earnings.

Habbash et al. (2010) examine the commitment of non-executive directors (i.e.

composition, meetings and directors fees) and chairman independence, using a sample

of 227 UK FTSE 350 firms for fiscal year 2005 and 2006. In addition, they also

control the number of board meeting in their earnings management model. They find

that the non-executive directors‟ commitment and chairman independence are

important to constrain the opportunistic earnings. Sun et al. (2010) investigate the

association between the corporate environmental disclosure and earnings management

and the interaction effect of the corporate governance mechanisms (i.e. size of boards

and audit committee meetings) on the relationship between earnings and corporate

environmental disclosure. They find no significant evidence the corporate

environmental disclosure to be linked with reduced earnings manipulation but suggest

that the audit committee diligence affect the relationship between earnings and

corporate environmental disclosure. Both studies, however, do not fully integrate the

other effective characteristics of board and audit committee in their model

specification.

Habbash (2010) examines more comprehensively the characteristics of corporate

governance and auditor quality including the size, proportion of independent

members, levels of financial expertise, number of meetings, audit fees, NAS fees and

the use of industry specialist auditors in the data from FTSE 350 between 2003 and

2006. The audit fees are measured using the natural log of audit fees and the ratio of

audit fees to total fees, while the NAS fees are defined as the natural log of NAS fees

and the ratio of audit fees to total fees. The measurement of the use of industry

specialist auditors is based on the market shares of audit fees and the number of audit

clients in a specific industry. The analyses are undertaken using two separate earnings

management models. In the first model, he examines the effect of board of director

characteristics on earnings management. The results suggest a negative relationship of

the size and the independence of boards to earnings management. In the second

model, he investigates the effect of audit committee characteristics and external

auditor variables on opportunistic earnings. He finds that the audit committees that are

comprised of higher proportion of independent members and equipped with financial

expertise are more likely to constrain earnings management. Moreover, the results

suggest that firms that pay higher audit fees and lower NAS fees and that have

employed industry specialist auditors are more likely to limit earnings manipulation

behaviour.

Although Habbash (2010) examines relatively similar characteristics of board, audit

committee and auditor quality variables to those being investigated in this thesis, he

does not incorporate both board and audit committee characteristics in a single model.

Prior studies suggest that the effectiveness of an audit committee is associated with

the composition of the entire board, therefore the joint monitoring roles of boards and

audit committees are likely to strengthen the firm‟s corporate governance overall

(Menon and William, 1994; Collier and Gregory, 1999; Cohen et al., 2002; Boo and

Sharma, 2008). In addition, the segregation of tests on audit committee characteristics

from those of board of director characteristics may result in an incomplete analysis of

earnings management.

In a similar vein, Klein (2002) investigates the effect of independent boards and audit

committees on earnings management using data from 692 US firms in the period

between 1992 and 1993. Her findings suggest that independent boards and

independent audit committees are effective in constraining opportunistic earnings. Xie

et al. (2003) investigate more board and audit committee characteristics, including the

level of CEO duality, board meeting frequency, the independence of board members,

the level of expertise, board size, audit committee size, audit committee meeting

frequency and audit committee expertise and independence in 282 US firms. They

employ the Jones model and Teoh et al.‟s (1998) model to detect earnings

management. They find that the firms with boards and audit committees with more

independent members, who are equipped with corporate or financial expertise, and

who have frequent meetings, experience lower levels of discretionary accruals.

Bédard et al. (2004) examine the relationship between audit committee characteristics

(e.g. size, independence, expertise and number of meetings) and earnings

management in a sample of 300 US firms which consists of two subsamples: 200

firms with aggressive earnings management (100 firms with the largest positive and

100 firms with largest negative discretionary accruals) and the firms with the lowest

levels of discretionary accruals around zero (50 positive and 50 negative). They find

that financial expertise and the presence of solely independent audit committees are

negatively related to the likelihood of aggressive earnings management. This is

consistent with the arguments that these characteristics improve their oversight

function in monitoring earnings management. The size and the meeting frequency of

audit committees are not significantly related to aggressive earnings management.

Vafeas (2005) analyses the data of 252 US firms in between1994 and 2000 to

examine several board of directors and audit committee characteristics on the quality

of reported earnings. The poor earnings quality is surrogated by small earning

increases and the negative earnings avoidance. Evidence suggests that the small

earnings increases are associated to the audit committee with insiders‟ directors, audit

committee with director who are business executive and less frequent of audit

committee meeting. While the results from the model of negative earnings avoidance

suggest no significant relationship between the likelihood of avoiding negative

earnings and the board and audit committee characteristics.41

Specifically, Vafeas (2005: 1104) examines audit committee characteristics such as the percentage of

committee insiders, percentage of active business executives, percentage of members with other audit

committee experience, audit committee size, audit committee meeting frequency, median stock

Chtourou et al. (2001) examine the effect of board and audit committee characteristics

(e.g. board size, board independence, board expertise, the presence of multiple

directorships and director‟s tenure, CEO duality, the presence of independent

nomination committees or solely independent audit committees, audit committee

expertise and audit committee meeting frequency) and the Big 6 auditors on the extent

of earning management in US firms. Their results suggest that audit committees with

a higher percentage of independent members, solely independent and meet at least

twice a year indicate a lower incidence of income-increasing accruals, while audit

committees with financial expertise reduce the incidence of income-decreasing

accruals. In addition, the results also suggest less likelihood of income-increasing

accruals when firms have a higher proportion of independent members and less

likelihood of income-decreasing accruals when the boards are larger in size and when

the board members are equipped with board‟ experience. However, no evidence

suggests that the employment of a Big 6 auditor constrains earnings management

activities.

In contradiction to Chtourou et al.‟s study on the impact of Big 6 auditors on earnings

management, Becker et al. (1998) suggest that the firms with Big 6 auditors are more

likely to report lower discretionary accruals than the firms with non-Big 6 auditors,

and this is consistent with the argument that the Big 6 auditors provide higher quality

audits and thus are more likely to constrain opportunistic earnings. Francis et al.

(1999) suggest that the firms with a greater tendency for accruals are more likely to

hire the Big 6 auditors as a credible signal to outsiders that they are less likely to

manipulate earnings. They further argue that, even though their results suggest that

the firms with Big 6 auditors have relatively higher levels of total accruals, they are

less likely to be associated with higher discretionary accruals.

Krishnan (2003b) investigates the pricing of discretionary accruals of firms audited by

Big 6 versus non-Big 6 auditors. His findings suggest that the relationship between

stock returns and discretionary accruals are stronger for the firms that are audited by

ownership of committee members, mean tenure per committee member, mean directorships per

committee member and mean committee memberships per committee member. For board of director

characteristics, the percentage of outsider directors and board size is included.

Big 6 auditors as compared to those using non-Big 6 auditors. This supports the

argument that Big 6 auditors improve the credibility of reported accruals. He also

claims that the association between discretionary accruals and future profitability is

greater for firms with Big 6 auditors, suggesting that the Big 6 auditors enhance the

ability of discretionary accruals to predict future profitability. Kim et al. (2003)

examine the direction of discretionary accruals and the selection of Big 6 auditors.

They suggest that the Big 6 auditors are more effective in constraining income

increasing accruals but less effective in constraining income decreasing accruals.

In respect to industry specialist auditors and earnings management, Krishnan (2003a)

investigates 4,422 US firms, audited by Big 6 auditors, in between 1989 and 1998.

The use of industry specialist auditors is measured using market share and portfolio

shares approaches. The cross-sectional Jones model is employed to detect earnings

management. He suggests that the firms with industry specialist auditors are more

efficient in mitigating opportunistic earnings than the firms with non-specialist

auditors. Balsam et al. (2003) examine 19,091 US firms at the financial year-ends for

1991 to 1999. They suggest that the levels of discretionary accruals are lower and that

earnings response coefficients are higher for the firms with industry specialist auditors

(and vice-versa).

Frankel et al. (2002) examine data from 3,074 proxy statements files with SEC

between 5, 2001 and June 15, 2001. They employ three measures of auditor

independence: the ratio of NAS fees to total fees; the percentile rank of audit fees and

NAS fees disaggregated by auditor; and the percentile rank of the sum total of audit

and NAS fees. The level of earnings management is estimated using the propensity to

just meet or beat earnings benchmarks, and the cross-sectional modified Jones model

is used to estimate the levels of discretionary accruals. They find that the firms with a

higher NAS fee ratio and a higher percentile rank of NAS fees are more likely to meet

or beat earnings benchmarks and report higher discretionary accruals. These results

are robust to alternative earnings management measures such as discretionary total

accruals, discretionary working capital accruals, performance-matched discretionary

accruals and also being applicable to the samples of income-increasing and income-

decreasing discretionary accruals.

In addition to their primary analysis of NAS fees, Frankel et al. (2002) also find that

firms with higher audit fees (measured by the percentile rank of audit fees) are likely

to have lower earnings management. This result supports the argument that higher

audit fees may probably increase the number of audit hours, resulting in higher audit

efforts (Deis and Giroux, 1996). It seems that higher auditor efforts may compensate

for illegal behavior, including earnings manipulation, because management are more

worried that such action may be discovered by the auditors. In agreement with this

proposition, Caramanis and Lennox (2007) find that audit hours are negatively related

to income-increasing discretionary accruals to meet earnings benchmarks. They

conclude that a low audit effort increases the likelihood of a manager manipulating

reported earnings.

Ashbaugh et al. (2003) replicated Frankel et al.‟s (2002) study using similar datasets

and tests. Specifically, they measure auditor independence using (1) the natural log of

audit fees, (2) the natural log of NAS fees, (3) the natural log of the sum of audit and

NAF fees and (4) the ratio of NAS fees to total fee. Earnings management is

measured using the models of performance-adjusted discretionary accruals (using the

portfolio technique and adding the ROA variable in the discretionary accrual

regression) and the earnings benchmark. Their findings are relatively similar to those

reported in Frankel et al.‟s study except that they find no evidence that associates

income-increasing discretionary accruals with the NAS fees ratio when performance-

adjusted measures are employed.

In other studies, Chung and Kallapur (2003) investigate a sample of 1,871 US firms,

audited by Big 5 auditors. They measure auditor independence as a ratio of the total

fees (audit and NAS fees) to the audit firm‟s US revenues. The discretionary accruals

are estimated using Jones‟ model. Their results fail to find any significant relationship

between NAS fees and discretionary accruals. In a similar vein, Ruddock et al. (2006)

examine the relationship between NAS and earnings conservatism. They also find no

evidence that higher levels of NAS are associated with reduced conservatism.

Larker and Richardson (2004) provide mixed results on the relationship between NAS

fees and earnings management. Consistent with Frankel et al. (2002) they find a

positive relationship between the ratio of NAS fees to total fees and earnings

management. Using other NAS fee measurements, they find that the firms with the

higher levels of audit and NAS fees are likely to have lower earnings management. In

their study, the auditor independence is measured using the ratio of NAS to total fees,

the NAS fees, the sum of audit and NAS fees and the abnormal fees measurement.

The cross-sectional Jones and modified Jones models are used to detect earnings

management. They conclude that the monitoring roles of auditors are dependent on a

firm‟s corporate governance structures.

Based on a test of 434 listed Australian firms, Davidson et al. (2005) argue that the

strength of internal governance mechanisms (e.g. board of directors, audit committee,

internal audit and external auditor) shapes the practice of earnings management. They

find that having an independent board, CEO duality and an independent audit

committee are significantly and negatively related to reduce levels of earnings

management. However, there is no evidence that internal audits, audit committee size,

audit committee meeting frequency and use of a Big 5 auditor are associated with the

level of earnings management.

Osma and Noguer (2007) document evidence from Spain on the monitoring

effectiveness of a board and its committees in relation to earnings management.

Specifically, they examine the existence of institutional directors, independent boards,

independent boards with financial expertise and independent audit committees and

independent nomination committees. Their findings suggest that the institutional

directors is negatively related to earnings management and that there is no evidence to

support that the boards and it committees are associated with earnings manipulation.

This contrasts with prior evidence documented in the US, UK and Australia. These

results suggest that institutional directors are more effective in constraining earnings

management practices than boards and their committees.

Park and Shin (2004) investigate the monitoring roles of boards and the practice of

earnings management in Canada. In similarity to Osma and Noguer (2007), their

results suggest that financial intermediaries and institutional directors play a major

role in constraining earnings manipulation and that there is no significant evidence

that outside directors and their tenures are associated with the incidence of earnings

management.

Overall, the results on the relationships of corporate governance characteristics and

auditor variables to constraining earnings management suggest mixed findings.

Failure to control corporate governance variables or auditor variables in a single

model may explain the conflicting results in prior studies as being due to incomplete

analyses of earnings quality determinants and the monitoring role of auditors, which

vary depending the strength of the client‟s corporate governance (Larcker and

Richardson (2004). Therefore, in this thesis, the investigation of corporate governance

characteristics and earnings management will incorporate auditor variables in order to

avoid this misspecification.

The evidence reviewed suggests that an effective board and audit committee and a

higher auditor quality are associated with a greater extent of monitoring functions, and

are thus likely to constrain opportunistic earnings. In conjunction with the evidence

from the prior literature and the arguments developed under Section 3.4 and 3.5

regarding the effectiveness of boards and audit committees, the present study

hypothesizes that boards of directors which are smaller in size, have more

independent members, possess financial expertise and have more frequent board

meetings, as well as audit committees which are larger in size, solely independent,

equipped with financial expertise and meet frequently, all lead to more effective

monitoring. These characteristics of board and audit committee are expected to

constrain opportunistic earnings. Stated differently, the present study tests the

following hypotheses:

H25: There is a positive relationship between the board’s size and earnings

management.

H26: There is a negative relationship between the independent board and earnings

management.

H27: There is a negative relationship between the board’s financial expertise and

H28: There is a negative relationship between the board’s meeting frequency and

H29: There is a negative relationship between the audit committee’s size and

H30: There is a negative relationship between the solely independent audit

committee and earnings management.

H31: There is a negative relationship between the audit committee’s financial

expertise and earnings management.

H32: There is a negative relationship between the audit committee’s meeting

frequency and earnings management.

Similarly, consistent with the evidence reviewed and the theoretical proposition of

auditors‟ quality differentiation, the present study argues that the effectiveness of

audit services varies among auditors. The higher quality auditors are expected to

detect earnings management and questionable accounting practices better than the

lower quality auditors. In this thesis, the higher quality of auditors is associated with

higher audit fees, lower NAS fees and the engagement of industry specialist auditors.

These expectations lead to the following hypotheses:

H33: There is a negative relationship between audit fees and earnings management.

H34: There is a positive relationship between NAS fees and earnings management.

H35: There is a negative relationship between industry specialist auditor and

3.10 Summary

In this thesis audit quality is defined as the technical ability of an auditor to detect

errors and their objectivity in reporting the discovered errors. Prior literature

recognises several proxies to measure audit quality including audit fees, NAS fees and

the use of industry specialist auditors. Consistent with the signalling or reputation

hypothesis, higher audit fees and the engagement of industry specialist auditors are

associated with higher auditor quality. While, the lower NAS is regarded as lower

auditor quality due to the scepticism of the regulators and the investors that higher

NAS could compromise the auditor independence. Evidence from prior studies

suggests that the boards of directors which are smaller in size, have more independent

directors, are equipped with financial expertise and meet more frequently are effective

in their monitoring role. Similarly, audit committees with more members, sole

independence, more financial expertise and that are more active are suggested to have

a higher oversight function. Therefore, consistent with agency theory proposition and

prior empirical evidence, the present study hypothesises that these effective

characteristics of boards and audit committees are associated with a higher audit

quality. In regard to earnings management, this thesis views earnings management as

opportunistic earnings. The present study argues that the firms with effective

characteristics of board and audit committee and higher quality auditors are less likely

to allow earnings management because opportunistic earnings cause uncertainty about

the economic value of a firm. Table 3.3 provides the summary of the key literature

covered in this chapter.

Table 3.1: Summary of the key literature relating to board of directors, audit committee, audit quality and earnings management studies

Author(s) Sample Country Audit quality

related proxy(s)

and earnings

management/

Dependent

variable

Board of directors

(BOD) and audit

committee (AC)

characteristics/

Independent

variable(s)

Results

Abbott et al.

(2003a)

492 non- regulated

firm and audited by

Big-5 auditors that

filed proxy

statement with

SEC in the period

February 5, 2001 to

June 30, 2001.

US Audit fees BOD – composition,

multiple directorship

AC – composition,

expertise, meeting

Higher audit fees are associated with:

1.solely independent non-executive directors in

audit committee

2.at least one member of audit committee

equipped with accounting or financial expertise

3. higher board meeting frequency

4. higher percentage of independent non-

executive directors on board

The audit committee meeting frequency and

board‟s expertise are not significantly correlated

with audit fees.

Carcello et al.

(2002)

258 firms from

Fortune 1000 for

fiscal year ended

between April

1992 and March

US Audit fees BOD – composition,

expertise, meeting

AC – composition,

expertise, meeting

The firms with the higher percentage of

independent non-executive directors on boards,

more expertise (multiple directorships) and

higher board meeting frequencies are likely to

have higher audit fees. When they replace the

board attributes with audit committee attributes

(i.e. composition, meeting frequency and

expertise), their results show a positive

relationship of audit committee independence and

audit committee expertise to audit fees. The audit

committee meeting frequency is not significantly

related with audit fees. However, once they

analyse board and audit committee attributes

together, the results for the board of directors

remain unchanged but none of the audit

committee attributes are significantly related to

audit fees.

Table 3.1 (continued)

Collier and

Gregory

(1996)

315 firms for year

1991 (Financial

Times All Share

Index)

UK Audit fees AC – establishment

The firms with the presence of an audit

committee are likely to have higher audit fees.

Tsui et al.

(2001)

650 firms from

1994 to 1996

Audit fees BOD – duality roles The firms that separate the functions of CEO and

chairman roles are more likely to have lower

audit fees, indicating effective monitoring

mechanisms are in place, thus reducing the

control risk and audit effort.

Boo and

Sharma, (2008)

469 firms with

total assets exceed

$US 1bilion in

fiscal year 2001

US Audit fees BOD/AC – size,

composition, expertise,

meeting

The associations of the board/audit committee

independence and board/audit committee size to

audit fees are weaker for regulated firms than for

non-regulated firms except in the case of

board/audit committee directorships.

O‟Sullivan

(2000)

402 UK large firms

in 1992

UK Audit fees BOD – composition,

ownership

The firms with the higher percentage of non-

executive directors on board are more likely to

incur higher audit fees, indicating a higher audit

quality. He also claims that there is a negative

relationship between executive ownership and

audit fees.

O‟Sullivan

(1999)

146 UK large firms

in 1995

tenure, duality roles

AC – size, composition

There is no evidence that the board of director or

audit committee attributes influence the level of

audit fees.

O‟Sullivan and

Diacon (2002)

117 insurance

companies in 1992

duality, ownership

AC – establishment,

composition

The existence of an audit committee has a

positive relationship with audit fees but the audit

committee composition as well as the board

characteristics has no significant relationship

with audit fees.

Krishnan and

Visvanathan,

(2009)

801 listed on the

S&P 500 for 2000

to 2002, audited by

Big 5 auditors.

US Audit fees BOD – size,

composition, meeting,

duality roles, voting

control, ownership

AC – composition,

expertise, meetings

The firms with separated dual roles functions and

audit committees equipped with financial

expertise are perceived by auditors to have a

strong internal control environment, thus

reducing the control risk and audit effort and

leading to lower audit fees. Board and audit

committee meetings have a positive relationship

with audit fees.

Abbott et al.

(2003b)

538 firms that filed

with SEC between

February 5, 2001

and March 15,

US Non-audit fees

(ratio non-audit

fees to total fees)

AC – composition,

meeting

BOD – ownership

structure

The firms that have audit committees that consist

of solely independent non-executive directors and

meet at least four times a year are likely to have a

lower NAS fees ratio.

Lee and Mande

(2005)

780 firms for fiscal

year 2000, S&P

Super 1500

US Non-Audit fees AC – composition,

expertise, meeting

The audit committees that are comprised of

solely independent non-executive directors and

that meet at least four times a year are likely to

limit the purchase of NAS. However, once they

model the NAS endogenously, such relationships

are insignificant.

Lee (2008) 631 firms for fiscal

year 2000 to 2001,

S&P Super 1500

US Changes in non-

audit fees ratio

BOD – expertise

AC – composition,

expertise

The independence and expertise of audit

committees and boards of directors are likely to

limit the level of NAS purchased.

Beasley and

Petroni (2001)

681 property-

liability insurers

during 1991 to

US Industry specialist

auditor

BOD - composition The insurers with the higher percentage of non-

executive directors are more likely to employ an

industry specialist auditor.

Chen et al.

(2005)

510 top firms listed

on the ASX in

Australia Industry specialist

auditor

BOD – multiple

directorship

AC – composition,

expertise, meetings

The firms with a higher percentage of non-

executive directors on their audit committees are

more likely to employ industry specialist

auditors. Audit committee expertise and the

number of meetings are not significantly related

to the engagement of industry specialist auditors.

The results for board directorships are mixed.

Abbott and

Parker (2000)

500 firms listed on

the NYSE, AMEX

or NASDAQ

exchanges in 1994

US Industry specialist

auditor

BOD –

composition,

ownership

AC – composition,

meeting

The firms with audit committees that are solely

comprised of independent non-executive

directors and that meet twice a year are more

likely to employ industry specialist auditors. The

percentage of non-executive directors on boards

is not significantly related to the engagement of

industry specialist auditor across all

measurements.

Dechow et al.

(1996)

92 firms that

subject to the SEC

enforcement action

between 1982 and

US Earnings management:

discretionary accruals

BOD –

composition,

ownership, duality

AC –

establishment

Where a firm‟s board of directors is dominated

by the management, members practice dual role

functions, the CEO is also the founder of the

firm, there are fewer representatives of outside

block holders and there is no audit committee

formation there is more likelihood of earnings

manipulation.

Klein (2002) 692 firms-years

listed on the S&P

500 for 1992 to

US Earnings management

: absolute value of

discretionary accruals

BOD –

composition,

ownership

AC – composition

Higher percentages of independent non-executive

directors on audit committee and on board are

associated with lower earnings management.

However, there is no significant relationship

between solely independent non-executive

directors and earnings management.

Xie at al

(2003)

282 firms-years

listed on the S&P

500 for 1992, 1994

and 1996

US Earnings

management:

current

discretionary

accruals

BOD – duality roles,

meetings,

composition, size,

expertise

AC – size,

composition,

expertise, meeting

Earnings management is less likely to occur in

firms whose board and audit committee are

equipped with a corporate and financial

background and have a higher percentage of

independent non-executive directors as well as

higher number of meetings.

Bédard et al.

(2004)

200 firms

(aggressive

earnings

management-

highest and lowest

income increasing/

decreasing ),

Compustat in 1996

US Earnings

management:

discretionary

accruals

BOD – multiple

directorship,

composition,

ownership

AC – composition,

expertise, size,

meeting

The firms with solely independent audit

committees that are also equipped with financial

expertise are less likely to have aggressive

earnings management. There is no significant

relationship of audit committee size and the

number of meeting to earnings management.

Davidson et al.

(2005)

434 listed

Australian firms

for financial year

ending 2000

Australia Earnings

management:

absolute value of

discretionary

accruals

BOD – composition,

duality roles

AC – composition,

size, meeting

Firms with a majority of independent non-

executive directors on boards and with solely

independent audit committees are associated with

lower earnings management.

Peasnell et al.

(2005)

1,271 firms for

fiscal year 1993 to

UK Earnings

management:

discretionary

accruals

BOD – size,

composition,

ownership, duality

AC - establishment

Firms with a higher percentage of non-executive

directors on board are associated with lower

income-increasing earnings management.

However, there is no evidence that the presence of

an audit committee affects the extent of income-

increasing earnings management.

Park and Shin

(2004)

539 firm-year for

1991 to 1997

Canada Earnings

management:

discretionary

accruals

ownership

There is no significant relationship between the

number of non-executive directors on board and

earnings management. However, the

representatives of active institutional shareholders

reduce earnings management.

Agrawal and

Chadha (2005)

159 matched-pair

of public firms that

restated earnings in

200 and 2001

US Restatement of

earnings

expertise, ownership,

duality roles

AC – composition,

expertise

The likelihood of earnings restatement is lower in

the companies whose board or audit committee has

independent non-executive directors with financial

expertise, but it is higher in companies in which

the CEO belongs to the founding family.

Carcello and

Neal (2000)

223 financial

distressed firms

during 1994

US Modified audit

report

AC – size,

composition

The higher the percentage of independent non-

executive directors on audit committees, the lower

the likelihood of the firms to receive a going-

concern audit opinion. The audit committee size is

not correlated with the likelihood of a going-

concern audit report.

McMullen

(1996)

219 of firms

consist of firms

that associates to

the litigation,

earnings

restatement, SEC

actions, illegal acts

and auditor

turnover with

disagreement

US Financial reporting

consequences-

Litigation,

earnings

restatement, SEC

actions, Illegal

Acts, Auditor

turnover

AC - establishment The existence of an audit committee is associated

with fewer lawsuits for fraud, fewer quarterly

earnings restatements, fewer SEC enforcement

actions, fewer illegal acts and fewer auditor

turnovers that are related to disagreements.

Chen et al.

(2006)

169 firms under the

Chinese Securities

Regulatory

Commission

(CSRC)

enforcement

actions during

1999 to 2003.

China Fraud and non-

fraud firms: coded

1 if the firm is

subject to an

enforcement action

statement, 0

otherwise

BOD- size,

composition,

meetings, duality

roles, chairman

tenure, ownership

The firms with the lower percentage of non-

executive directors on board, lower board meeting

frequencies, and shorter chairman tenures are

associated with a higher likelihood of the incidence

of fraud.

Abbott et al.

(2004)

1. match-paired of

88 restatement

sample firms from

1991 to 1999

2. 44 fraud firms

under the SEC

sanctions

US 1.Restatement

2.Fraud and non-

fraud firms

BOD – size, composition,

ownership, duality roles

AC – size, composition,

expertise, meeting

The firms with audit committees that are

solely independent, meet frequently and

that possess at least one member with

financial expertise are less likely to

experience restatement. A larger board size

is associated to a higher likelihood of

restatements. Audit committee expertise

and independence are negatively related to

the incidence of fraud.

Abbott et al.

(2000)

156 firms: 78 firms

subject to SEC‟

sanction matched

with 78 non-

sanctioned firms

US Sanctioned and non-

sanctioned firms:

coded 1 if the firm is

alleged to have SEC

sanction, 0 otherwise

AC – composition,

meeting

The firms with audit committees that are

comprised of solely independent members

and that meet twice a year are less likely to

be sanctioned for fraud and aggressive

accounting. Having a CEO who also chairs

the board is associated with a higher

likelihood of sanction.

Beasley (1996) 150 firms:

Matched-pair of 75

fraud and 75 non-

fraud firms (SEC

and Wall Street

Journal Index)

US Fraud and non-fraud

firms: coded 1 if the

firm is alleged to

have fraudulent

financial statement,

0 otherwise

Non-fraud firms are likely to experience

lower levels of fraudulent financial

reporting when the board has higher

percentage of independent non-executive

directors as compared with fraud firms.

The board‟s composition rather than audit

committee‟s presence is more important in

reducing fraudulent financial reporting.

CHAPTER 4

METHODOLOGY

4.1 Introduction

This chapter presents the methodology used to test the hypotheses outlined in Chapter

3. The first section explains and justifies the sample firms selected and the time period

during which the investigation was carried out. The chapter then outlines the

definitions and measurements of the hypothesis variables (i.e. the effective

characteristics of board and audit committee, audit quality proxies and earnings

management). The model specifications and related control variables, the sources of

data and the data analysis procedures are also discussed. Finally, a summary of the

chapter‟s contents is also provided.

4.2 Sample firms and period of study

The sample population for this study is the Financial Times Stock Exchange (FTSE

hereafter) 350, which consists of FTSE 100 and FTSE 200, representing the largest

350 firms by market capitalisation listed on the London Stock Exchange (LSE). These

firms are selected because they include a broad range of industrial and commercial

activities and account for a significant portion of the UK economic output.

The main focus of this thesis is to examine the effectiveness of the board of directors

and audit committee in terms of corporate governance best practice, as outlined in the

UK Corporate Governance Code (2010). Since some of the Code‟s provisions do not

apply to the firms below the FTSE 350, the FTSE 350 firms provide an ideal sample

to investigate (The UK Corporate Governance Code, 2010: 5).

The study examines a sample period of the fiscal years 2005 and 2008, following an

event believed to have significant influence on this study. The event in question is the

revision of the UK Corporate Governance Code, which was introduced in July 2003

based on the recommendations set out in the Higgs and Smith reports, and which

came into effect on 1 November 2003 for all UK listed firms. The Higgs Report

provides a review of the role and the effectiveness of non-executive directors, while

the Smith Report details the role of the audit committee. Allowing for a year‟s

transition period, the present study examines the fiscal period of 2005 to 2008 to

ensure that all FTSE 350 firms meet the new requirements.

4.2.1 Sample selection for regression analysis

The initial sample of the FTSE 350 consists of 1,400 firm-year observations for the

period 2005 to 2008. However, the present study excludes 504 firms that operate in

the financial and utilities sectors, due to their unique characteristics and to specific

regulations which may affect the results. The sample size has also been reduced by a

further 168 firm-year observations as a result of missing information in Datastream,

FAME and Thomson One Banker. Another 49 firm-year observations have been

eliminated, since their annual reports were unavailable, as have another 5 firms which

were not audited by Big 4 auditors in order to control for brand name (Craswell et al.,

1995; Chung and Kallapur, 2003). After the eliminations, the remaining sample is of

674 firm-year observations. The sample selection procedure is summarised in Table

Table 4.1: Sample selection procedures

Description 2005 2006 2007 2008 Pooled

Initial sample (FTSE 350) 350 350 350 350 1,400

Excluded:

Financial and Utilities firms 146 120 115 123 504

Missing data from Datasteam,

FAME and Thomson One Banker 31 39 46 52 168

Audited by non-Big 4 auditors 1 2 1 1 5

Unavailable annual report 5 8 7 29 49

Final samples 167 181 181 145 674

Two empirical investigations have been conducted, each of which use different

samples. The first investigation examines the relationship between the board of

directors, the audit committee and audit quality. The total sample analysed through

the three proxies of audit quality is 674 firm-year observations.

The second empirical investigation examines the relationship between the board of

directors, the audit committee and auditor quality in constraining earnings

management. In line with the arguments set forth by DeFond and Jiambalvo (1994)

and Subramanyam (1996), the sample for the second investigation is reduced from

674 to 613 to ensure that each industry portfolio consists of at least six observations,

in order to provide an unbiased estimation of discretionary accruals. Table 4.2 and

Table 4.3, Panel A and B, report the distribution of the sample firms by year and

industry, for the first and second empirical models respectively.

Table 4.2: Sample size and industry description for the first empirical analysis –

relationships between board of directors, audit committee and audit quality

Panel A: Distribution of sample firms by year

Year 2005 2006 2007 2008 Pooled

Sample

N Percent N Percent N Percent N Percent N Percent

167 24.78 181 26.85 181 26.85 145 21.52 674 100.00

Panel B: Distribution of sample firms by industry

Code Supersector level N Percent

0500 Oil & Gas 36 5.34

1300 Chemicals 16 2.38

1700 Basic Resources 30 4.45

2300 Construction & Materials 20 2.97

2700 Industrial Goods & Services 229 33.98

3500 Food & Beverage 23 3.41

3700 Personal & Household Goods 49 7.27

4500 Health Care 18 2.67

5300 Retail 74 10.98

5500 Media 47 6.97

5700 Travel & Leisure 77 11.42

6500 Telecommunication 17 2.52

9500 Technology 38 5.64

Total 674 100.00

The sample observations consist of FTSE350 non-regulated firms audited by Big 4

auditors.

Table 4.3: Sample size and industry description for the second empirical analysis –

relationship between board of directors, audit committee, external auditor and

earnings management

Panel A: Distribution of sample firms by year

Year 2005 2006 2007 2008 Pooled

Sample

N Percent N Percent N Percent N Percent N Percent

149 24.31 167 27.24 170 27.73 127 20.72 613 100.00

Panel B: Distribution of sample firms by industry

Code Supersector level N Percent

0500 Oil & Gas 36 5.87

1300 Chemicals - -

1700 Basic Resources 26 4.24

2300 Construction & Materials 18 2.94

2700 Industrial Goods & Services 229 37.36

3500 Food & Beverage 19 3.10

3700 Personal & Household Goods 49 7.99

4500 Health Care - -

5300 Retail 74 12.07

5500 Media 47 7.67

5700 Travel & Leisure 77 12.56

6500 Telecommunication - -

9500 Technology 38 6.20

Total 613 100.00

From the total sample of 674 firm-year observations, the sample has been reduced to

613 firm-year observations to ensure that each industry portfolio consists of at least

six observations in every year.

4.3 Sources of data

There are four main sources of data relevant to the study, namely Datastream, FAME,

Thomson One Banker and the firms‟ annual reports. Most of the variables identified in

this study were taken from Datastream, FAME and Thomson One Banker. The other

variables, for which data was not available from these sources, were collected from

the individual firms‟ annual reports; this refers in particular to the variables relating to

the board of directors and the audit committees.

4.4 The definition and measurement of the hypothesis variables

The variables of interest examined in this study are explained in this section. There

are three main variables to be examined: (1) the characteristics of the board of

directors and the audit committee; (2) the audit quality proxies; and (3) earnings

management.

4.4.1 The board of director and audit committee variables

The board of director and audit committee variables are related to their effective

monitoring characteristics (e.g. size, composition, meeting and expertise). In both

empirical investigations, the board and audit committee variables are independent

variables. These variables are manually collected from the firms‟ annual reports, and

each variable is presented in the following subsections.

4.4.1.1 Board of directors size (BRDSIZE)

Following Abbott et al. (2004) and Peasnell et al. (2000; 2005), the board of directors‟

size is determined by the total number of members on the board of directors, as

presented in the annual report at the end of each fiscal year.

4.4.1.2 Board of directors composition (BRDNED)

The board of directors‟ composition is defined as the proportion of independent non-

executive directors to total board size. Non-executive directors can be either

independent non-executive directors or gray or affiliated non-executive directors.

However, the independent non-executive directors are believed to have a better

monitoring position compared to the gray or affiliated non-executive directors, since

they have no relationship that would impair their judgments or decision-making

(Lawrence and Stapledon, 1999; Vance, 1983). This is consistent with agency theory

(Jensen and Meckling, 1976).

The definition of „independent‟ is based on the UK Corporate Governance Code

(2010: 8), paragraph A.3.1: “The non-executive director is not considered to be

independent if the director:

(1) has been an employee of the company or group within the last five years;

(2) has, or has had within the last three years, a material business relationship with

the company either directly, or as a partner, shareholder, director or senior

employee of a body that has such a relationship with the company;

(3) has received or receives additional remuneration from the company apart from a

director‟s fee, participates in the company‟s share option or a performance-

related pay scheme, or is a member of the company‟s pension scheme;

(4) has close family ties with any of the company‟s advisers, directors or senior

employees;

(5) holds cross-directorships or has significant links with other directors through

involvement in other companies or bodies;

(6) represents a significant shareholder; or

(7) has served on the board for more than nine years from the date of their first

election.”

The independent non-executive directors exclude the chairman of the board.

4.4.1.3 Board of directors expertise (BRDEXP)

The board of directors‟ expertise is measured by the number of directors with

accounting and financial qualification and experience, in proportion to the total board

size. Accounting and financial expertise provide the board members with an

understanding of financial statements, which enables them to assess the effectiveness

of the accounting policies provided by the management. Accounting and financial

qualifications and experience include all relevant forms of formal education (e.g.

Bachelor in Accounting) and professional qualification (e.g. ACCA, CIMA and

CFA), as well as work experience (as finance director, chief financial officer, auditor

or financial controller, for example). This definition of the variable is relatively

similar to Xie et al. (2003).

4.4.1.4 Board of directors meeting (BRDMEET)

Following Vafeas (1999), the board of directors‟ meetings are measured by the total

number of board of directors‟ meetings during the year. The meetings frequency

captured the boards‟ activity levels. Regular board meetings allow the board members

to identify and resolve potential problems that may arise in their firm, particularly

related to its financial health. The meetings are also an opportunity to generate and

analyse strategic planning.

4.4.1.5 Audit committee size (ACSIZE)

The audit committee size is measured based on the total number of audit committee

members presented at the end of fiscal year; this definition is consistent with Yang

and Krishnan (2005).

4.4.1.6 Audit committee composition (ACIND)

The audit committee composition is measured by a dummy variable, coded as 1 if the

audit committee consists solely of independent directors and coded as 0 if otherwise,

consistent with the recommendations outlined in the UK Corporate Governance Code

(2010). The rationale for having solely independent non-executive directors in the

audit committee is to ensure greater impartiality and objectivity in decision-making.

The definition of independence is relatively similar as defined in section 4.4.1.4.

4.4.1.7 Audit committee expertise (ACEXP)

As with the board of directors‟ expertise, audit committee expertise is measured by

the number of directors with accounting and financial qualification and experience, in

proportion to the total number of audit committee members. Accounting and financial

expertise is essential to the audit committee members, since their main function is to

ensure integrity in financial reporting. The accounting and financial knowledge may

increase their understanding of financial reporting issues and of GAAP practices as

well as better enabling them to assess the effectiveness of the firms‟ accounting

policies. Accounting and financial qualifications and experience include all forms of

formal education, professional qualification and work experience related to

accounting and finance.

4.4.1.8 Audit committee meeting (ACMEET)

Following Krishnan and Visvanathan, (2009), the audit committee meeting is

measured by the number of meeting during the fiscal year. The higher the number of

the meetings, the higher the level of activities, which mean the more active they are.

4.4.2 Audit quality variables

Three proxies of audit quality are examined in this thesis. In the first empirical

analysis, the audit quality proxies are dependent variables, whilst in the second, they

are represented as independent variables. The definition and measurement of each

audit quality proxy are described below.

4.4.2.1 Audit fees (LNAFEE)

The present study uses audit fees as a first measure of audit quality. The audit fees

variable (AFEE) is transformed to natural log and prefixed by LN to achieve normality

of data, in order to prevent the largest firms from unduly influencing the findings.

Data for this variable (LNAFEE) is gathered from the FAME database.

4.4.2.2 Auditor independence (FEE)

Four measures of auditor independence are employed; (1) LNNAF- the magnitude of

NAS fees; (2) LNTOTALFEES - the sum of audit and NAS fees; (3) FEERATIO1- the

fee ratio of NAS fees to total fees; and (4) FEERATIO2 - the fee ratio of non audit

fees to audit fees. The LNNAF and LNTOTALFEES are transformed to the natural log

to achieved normality of data distribution. The firms that reported zero NAS fees are

set to one pound to allow for log transmission. The auditor fees data were taken from

the FAME database.

LNNAF and LNTOTALFEES are associates to the fee dependence of auditor on a

client. Besides the audit fees received by the auditor, the level of NAS fees further

increase the client-auditor economic bond as the auditor reliance on the client

increases (Simunic, 1984; Beck et al., 1988). Similar to Ashbaugh et al. (2003), the

present study argues that the level of NAS fees and the sum of audit and NAS fees are

better measures to capture the economic important of the client to the auditor than the

NAS ratio. Although FEERATIO1 and FEERATIO2 do not necessarily capture the

client‟s importance, they do explain the financial link between the auditor and the

client, and have an impact on the regulators‟ perception of independence (Ashbaugh

et al., 2003).

4.4.2.3 Industry specialist auditor (SPEC_AUD)

The existing literature suggests that the industry specialisation of auditors can be

measured using several approaches, such as the market share approach (Krishnan,

2003a; Dunn et al., 2000; Balsam et al., 2003; Velury et al., 2003; Chen et al., 2005)

and the portfolio approach (Krishnan, 2003a), as well as a complementary approach

set out by Neal and Riley (2004). Despite the limitations of each approach, they are

recognised as the most appropriate measures for auditor industry specialisation.

The market share approach interprets the industry specialist auditor as an auditor that

can make a distinction among their opponents within a particular industry in terms of

market shares (Neal and Riley, 2004). Market shares can be estimated for a specific

industry using client sales, the audit fees, the total fees and the number of audit clients

that are assigned to the particular auditor. The auditor(s) with the largest market

share(s) in a particular industry (within-industry) are assumed to have the largest

industry specific knowledge and a significant investment contribution towards the

improvement of audit quality and economic of scales as compared to others

competitors. Neal and Riley (2004: 170) claim that there are two disadvantages of

using the market shares approach in defining industry specialist auditors: (1) more

highly competitive auditors do not necessarily allocate significant resources to the

development of audit technology and expertise; and (2) it is not appropriate for

auditors to designate specialist in industries that are too small in generating significant

revenues for them.

An alternative to the market share approach, the portfolio approach assumed each

auditor individually and takes into consideration the distribution of the client‟s sales,

the audit fees, the total fees or the number of audit clients across the various

industries. The auditors with the largest portfolio share are considered as industry

specialist auditors in a particular industry if they generate the most revenues from

their total income or clients‟ sales. This may reflect their investments in audit

technologies and their industry-specific knowledge, even though they are not the

market share leaders in that particular industry (Neal and Riley, 2004). However, even

though the portfolio approach suggests that industry specialist auditors are driven by

the size of industry, the specific efforts or investment made by the auditors may not

necessarily reflect their expertise in that particular industry. This inconsistency may

cause larger auditors to be identified as specialists in most industries and none to be

identified as specialists in the smallest industries (Neal and Riley, 2004).

Neal and Riley (2004) introduce the weighted market share, a complementary

approach which captures the complementary effects between the market share

approach and the portfolio approach. This measure offers a solution for the

inconsistency between the two main approaches. However, Neal and Riley note that,

like the other two approaches, this approach does not consider the lead/lag period

effect. Thus, to ensure the consistency and robustness of the findings, the present

study considers all three of these approaches in determining the industry specialist

auditors based on audit fees.

In both empirical analyses, the industry specialist auditor is defined in five ways. The

first three measures are known as a continuous variable which is equal to the

respective auditors‟ market share (SPECLST_MS), the auditors‟ portfolio share

(SPECLST_PS) and compliment between the auditors‟ market share and portfolio

share (SPECLST_WEIGHTED). The last two measures are identified as dichotomous

variables that depend on the auditors‟ industry market shares; first is the industry

leader, SPECLST_MSLEADER, coded as 1 if the auditor has earned the largest market

share in each particular industry and 0 if otherwise, and lastly SPECLST_30MS, coded

as 1 if the auditor market share exceeds 30 percent in each particular industry and 0 if

otherwise. Each of these measures is computed as follows:

(1) SPECLST_MS

This is a continuous variable. SPECLST_MS is defined as the proportion of

audit fees received by the individual auditor relative to the total audit fees

received by all auditors in that particular industry (Velury et al., 2003).

ik I J

SPECLST MS

i = an index of the auditors (i=1, 2, 3, 4);

j = an index of clients;

k = an index of the industries

Ik = the number of auditors in industry k for year t;

Jikt = the number of clients audited by auditor i in industry k for

year t;

AFEEijkt = the audit fees for auditor i‟s client j for year t.

The variable AFEE denotes the client‟s audit fees and is sourced from the

FAME database. The numerator is the sum of the audit fees of all Jik clients of

auditor i in industry k, where the industry is defined at the sector level of the

Industry Classification Benchmark (ICB) available in Datastream. The

denominator in equation (1) is the audit fees of all Jikt clients in industry k and

year t reported in the FAME database, summed over all Ik auditors (all Big 4

auditors) providing audits to that particular industry and year.

(2) SPECLST_PS

This is a continuous variable. SPECLST_PS is identified as the proportion of

the total audit fees in each industry divided by the total audit fees of all

industry received by the particular auditor. The computation of the auditor

portfolio share is shown below:

SPECLST PS

i = an index of the auditors (i=1, 2, 3, 4);

j = an index of clients;

k = an index of the industries

Jikt = the number of clients audited by auditor i in industry k for

year t;

AFEEijkt = the audit fees for auditor i‟s client j for year t.

AFEE represents the client‟s audit fees and is sourced from the FAME

database. As with the market share approach, the numerator is the sum of the

audit fees of all Jik clients of auditor i in industry k, where the industry is

defined at the sector level of the ICB. The denominator is the sum of the audit

fees received by auditor i across all k industries in year t.

(3) SPECLST_WEIGHTED

The last measure is a complementary measure which incorporates both the

market share and the portfolio approaches. This is a continuous variable and

was introduced by Neal and Riley (2004). SPECLST_WEIGHTED is

computed as variable SPECLST_MS multiplied by SPECLST_PS.

(4) SPECLST_MSLEADER

This is a dichotomous variable. SPECLST_MSLEADER is coded as 1 if the

incumbent auditor earned the largest market share in the particular industry, 0

if otherwise. This measure identifies the auditor as the industry market leader.

The computation is the same as Equation (1).

(5) SPECLST_30MS

This measure is a dichotomous variable. SPECLST_30MS is coded as 1 if the

auditor market share exceeds 30 percent in each particular industry, 0 if

otherwise. The market share cut-off for specialisation is identified as at 30

percent without specialisation, with each firm holds a market share of

approximately 0.25 percent (1 firm/4 firms = 0.25). The 0.25 percent is

multiplying by 1.20 yields 30 percent (Neal and Riley, 2004). The

computation of the auditor industry market share is as in Equation (1).

Based on the computations of the market share and portfolio share approaches, Tables

4.4 and 4.5 report the summary of Big 4 auditor industry specialists by year and

pooled samples respectively. Across the years and pooled tables, according to the

SPECLST_30MS definition, it can be considered that PWC is the specialist in most

industries, while EY is considered to be a specialist only in the travel and leisure

industry, except in 2008, when it also was known as a specialist in the media and

technology industries. DL and KPMG are relatively comparable in terms of the

number of specialisations. For example, in the period 2005 to 2008 (pooled), DL is a

specialist in four industries: basic resources, construction & materials, media and

telecommunications, and KPMG is also a specialist in four industries, namely

chemicals, construction & materials, industrial goods & services and food &

beverages.

Table 4.4: The Big 4 auditor industry specialists (by year)

Panel A: Industry Market Share (in %)

2005 2006

DL EY KPMG PWC DL EY KPMG PWC

0500 23.856 8.106 - 68.038 10.269 14.463 - 75.268

1300 9.926 - 82.410 7.664 - - 87.237 12.763

1700 38.721 22.562 24.638 14.079 31.720 21.865 24.993 21.422

2300 38.423 - 29.287 32.29 41.855 - 58.145 -

2700 17.694 11.773 42.551 27.982 15.012 9.617 39.208 36.163

3500 4.545 4.545 60.606 30.304 3.333 5.000 33.333 58.334

3700 2.059 - 7.923 90.018 2.611 - 10.552 86.837

4500 11.429 14.286 7.143 67.142 2.684 12.567 28.199 56.550

5300 19.929 5.284 14.590 60.197 18.251 5.979 12.901 62.869

5500 49.239 7.711 6.425 36.625 69.483 4.594 3.281 22.642

5700 27.722 35.642 5.657 30.979 20.556 49.549 5.798 24.097

6500 34.458 - - 65.542 55.523 - 4.331 40.146

9500 8.299 22.189 3.239 66.273 28.619 3.773 4.463 63.145

2007 2008

0500 29.228 8.115 - 62.657 15.822 15.583 23.859 44.736

1300 - - 66.741 33.259 25.974 - 74.026 -

1700 18.761 23.195 16.314 41.730 48.985 - 48.985 2.030

2300 62.006 - 37.994 - 80.000 - 20.000 -

2700 20.172 11.015 35.617 33.196 21.358 4.131 42.441 32.070

3500 2.044 5.399 38.565 53.992 23.097 1.768 44.197 30.938

3700 7.286 - 7.764 84.950 3.435 - 6.304 90.261

4500 3.029 11.408 28.521 57.042 - 14.218 33.649 52.133

5300 18.301 5.396 10.402 65.901 24.924 2.935 7.340 64.801

5500 63.155 7.774 6.303 22.768 61.721 31.190 7.089 -

5700 22.326 48.421 6.316 22.937 30.679 39.424 7.755 22.142

6500 61.814 - 7.636 30.550 35.638 - 18.837 45.525

9500 25.941 7.377 - 66.682 26.904 30.002 1.454 41.640

Panel B: Industry Portfolio Share (in %)

2005 2006

0500 2.352 1.470 - 3.808 0.775 2.474 - 3.641

1300 1.072 - 8.458 0.470 - - 6.871 0.588

1700 22.776 24.413 13.765 4.701 13.587 21.216 11.726 5.877

2300 4.928 - 3.567 2.351 3.651 - 5.555 -

2700 23.507 28.777 53.693 21.102 17.427 25.291 49.854 26.886

3500 0.621 1.143 7.866 2.351 0.612 2.079 6.701 6.856

3700 0.725 - 2.649 17.987 0.612 - 2.707 13.026

4500 3.313 7.617 1.966 11.047 0.653 6.929 7.518 8.815

5300 5.311 2.590 3.693 9.106 4.529 3.361 3.506 9.991

5500 15.865 4.570 1.967 6.699 32.386 4.850 1.675 6.759

5700 10.144 23.994 1.967 6.436 5.965 32.570 1.843 4.478

6500 8.282 - - 8.941 15.685 - 1.340 7.263

9500 1.104 5.426 0.409 5.001 4.118 1.230 0.704 5.820

2007 2008

0500 2.913 1.745 - 3.824 1.700 4.718 2.318 3.185

1300 - - 2.165 0.521 0.960 - 2.478 -

1700 9.790 26.125 10.788 13.336 11.213 - 10.148 0.308

2300 7.236 - 5.619 - 6.408 - 1.450 -

2700 22.088 26.030 49.425 22.259 25.675 13.997 46.171 25.552

3500 0.376 2.144 8.991 6.083 8.372 1.806 14.497 7.432

3700 1.986 - 2.683 14.184 1.282 - 2.128 22.318

4500 0.753 6.125 8.991 8.689 - 13.545 10.293 11.679

5300 4.668 2.970 3.363 10.294 6.408 2.127 1.708 11.042

5500 16.137 4.287 2.041 3.563 13.948 19.866 1.450 -

5700 6.019 28.173 2.158 3.787 7.604 27.542 1.740 3.638

6500 24.121 - 3.776 7.301 11.213 - 5.364 9.494

9500 3.912 2.401 - 6.159 5.217 16.399 0.255 5.352

ICB Code (supersector levels): 0500-Oil & Gas; 1300- Chemicals; 1700- Basic

Resources; 2300- Construction & Materials; 2700- Industrial Goods & Services;

3500- Food & Beverage; 3700- Personal & Household Goods; 4500- Health Care;

5300- Retail; 5500- Media; 5700- Travel & Leisure; 6500- Telecommunications;

9500- Technology; DL: Deloitte; EY: Ernst & Young; KPMG: KPMG Peat Marwick;

PWC: Price Waterhouse Coopers. The audit fees is used as the based in calculating

the auditor industry expertise. The following examples illustrate how the auditor

industry expertise are been computed. For the period 2005, the total audit fees

received by DL in the Oil & Gas industry amounted £1.136 millions and the total

audit fees of DL for all industries amounted £48.3 millions. During the same period,

the combined audit fees of all auditors (DL, EY, KPMG and PWC) in the Oil & Gas

industry amounted £4.762 millions. The DL market share Oil & Industry,

2005=£1.136/£4.762*100=23.856%. The DL portfolio share Oil & Industry,

2005=£1.136/£48.3*100=2.352. The industry expertise for other auditors and the

subsequence years are also been computed in a similar ways. The auditor expertise in

bold is where market shares exceed 30%. A dash (-) means that the auditor did not

have client in that particular industry.

Table 4.5: The Big 4 auditor industry specialist (by pooled)

Auditor Industry Expertise for 2005-2008 (in %)

DL EY KPMG PWC

MS PS MS PS MS PS MS PS

0500 20.304 1.929 11.588 2.470 6.829 0.680 61.279 3.621

1300 7.966 0.453 0.000 0.000 80.633 4.812 11.401 0.404

1700 31.431 13.701 19.598 19.163 25.119 11.483 23.852 6.467

2300 55.199 5.624 0.000 0.000 36.840 3.936 7.961 0.504

2700 18.593 22.038 9.051 24.068 39.824 49.502 32.532 23.985

3500 11.424 2.511 3.692 1.819 42.442 9.781 42.442 5.801

3700 3.943 1.197 0.000 0.000 7.916 2.519 88.141 16.635

4500 3.732 1.037 13.130 8.187 25.658 7.480 57.480 9.938

5300 20.304 5.197 4.897 2.812 11.120 2.985 63.679 10.139

5500 62.593 19.838 10.748 7.642 5.291 1.759 21.368 4.212

5700 25.012 7.214 43.588 28.202 6.328 1.914 25.072 4.499

6500 49.539 15.514 0.000 0.000 8.547 2.807 41.914 8.164

9500 23.959 3.747 16.066 5.637 2.085 0.342 57.890 5.631

ICB Code (supersector levels): 0500-Oil & Gas; 1300- Chemicals; 1700- Basic

Resources; 2300- Construction & Materials; 2700- Industrial Goods & Services;

3500- Food & Beverage; 3700- Personal & Household Goods; 4500- Health Care;

5300- Retail; 5500- Media; 5700- Travel & Leisure; 6500- Telecommunications;

9500- Technology; DL: Deloitte; EY: Ernst & Young; KPMG: KPMG Peat Marwick;

PWC: Price Waterhouse Coopers; MS: market share, PS: Portfolio share; The audit

fees is used as the based in calculating the auditor industry expertise. The following

examples illustrate how the auditor industry expertise are been computed. For the

period 2005 through 2008, the total audit fees received by DL in the Oil & Gas

industry amounted £4.757 millions and the total audit fees of DL for all industries

amounted £246.59 milions. During the same period, the combined audit fees of all

auditors (DL, EY, KPMG and PWC) in the Oil & Gas industry amounted £23.429

millions. The DL market share Oil & Industry, 2005-2008=£4.757/£23.429*100=20.304%.

The DL portfolio share Oil & Industry, 2005-2008=£4.757/£246.59*100=1.929. The DL

weighted market share Oil & Industry, 2005-2008=20.304*1.929/100=0.392. The industry

expertise for other auditors are been computed in a similar ways. The auditor

expertise in bold is consider to be a specialist according to SPECLST_30MS’

definitions (i.e. the auditors are considered to be a specialist when they had the market

shares that exceed 30 percent).

4.4.3 Earnings management variables

Earnings management is the dependent variable in the second empirical analysis and

is measured by the absolute value of discretionary accruals. In accordance with

Becker et al. (1998), the absolute value of discretionary accruals measures the level of

opportunistic earnings management activities and extreme reporting decision

exercises by the managers.

The total accruals are identified in order to estimate the discretionary accruals. There

are two methods in computing the total accrual accruals; the first is the traditional

balance sheet approach (e.g. Healy, 1985; Dechow et al., 1995), and the second is the

cash flow approach (e.g. Becker et al., 1998; Subramaniam, 1996; Xie et al., 2003).42

Both approaches are extensively used in the prior literature. However, Hribar and

Collins (2002) suggest that the cash flow approach is better than the balance sheet

approach when estimating the accruals for earnings management. They claim that the

balance sheet approach contain the error measurement, which can lead to erroneously

conclusion of the existing of earnings management when no such earnings

management was detected. Following Hribar and Collins (2002) argument, the

present study employed the cash flow approach in computing the total accruals.

The discretionary accruals are estimated using a cross-sectional variation of the Jones

(1991), the modified Jones (1991) by Dechow et al. (1995) and the performance-

adjusted model by Kothari et al (2005). Following DeFond and Jiambalvo (1994), the

cross-sectional version is employed as it is more specific than the time version model

Under the balance sheet approach, the accruals are measured as follows (Hribar and Collins, 2002:

TACCt= ΔCAt – ΔCLt – ΔCasht + ΔDEBT – DEP

where TACC=total accruals; ΔCAt=the change in current assets during year t; ΔCLt=the change in

current liabilities during year t; ΔCasht=the change in cash and cash equivalent during period t;

ΔDEBT=the change in debt included in current liabilities during period t and DEP=depreciation and

amortization expenses during period t. All variables are scaled by lagged total assets.

The cash flow approach, on the other hand, measures the accruals as the difference between earnings

before extraordinary items and earnings before discontinued operation, less the operating cash flow

(Hribar and Collin, 202: 109).

due to the small sample observations (Subramanyam, 1996). All data is sourced are

from Datastream.

4.4.3.1 Discretionary accruals under the Jones model (DACCJM)

Two steps are involved in the estimation of discretionary accruals. First is the

estimation of non-discretionary accruals, using the following model. This begins with

the estimation of coefficients ά1, ά2, and ά3 for each industry in each year (at least 6

firms in each industry) by using OLS regression. The second step is to estimate the

error term in the model, which represents the discretionary component of accrual.

This error term is the difference between the total accruals and the non-discretionary

accruals.

1 1 1 1

1ijt ijt ijt

ijt ijt ijt ijt

TACC REV PPEά ά ά e

TA TA TA TA

where:

DACC ij = discretionary accruals for sample firm i in industry j for year t;

TACCijt = total accruals for sample firm i in industry j for year t;

TAijt -1 = total assets for sample firm i in industry j for year t-1;

ΔRECijt = change in account receivable for sample firm i in industry j for year t;

PPEijt = gross property plant and equipment for sample firm i in industry j for

year t;

eijt = error term for sample firm i in industry j for year t;

The total accruals are computed as earnings before extraordinary items and earnings

before discontinued operations, less the net cash flows from operating activities. The

industry (j) is classified using the Industry Classification Benchmark (ICB) by Dow

Jones and FTSE.

4.4.3.2 Discretionary accruals under the modified Jones model (DACCMJM)

The estimation of discretionary accruals under the modified Jones (1991) model is

relatively similar to the original Jones model, except that it takes into account the

changes in account receivable. The non-discretionary accruals are estimated using the

model below; the steps involved are relatively similar to the original Jones (1991)

model.

1 1 1 1

1ijt ijt ijt ijt

ijt ijt ijt ijt

TACC REV REC PPEά ά ά e

TA TA TA TA

where:

DACC ij = discretionary accruals for sample firm i in industry j for year t-1;

ΔREVijt = change in revenues for sample firm i in industry j for year t;

year t;

4.4.3.3 Performance-Adjusted Discretionary Accruals (DACCROA)

Kothari et al. (2005) suggest that there are two ways to control the firms‟ performance

in the estimated accruals. The first is by matching each firm-year observation with

another from a similar industry and year with the closest ROA in the current.

Alternatively, firm performance, including ROA, can be included in the discretionary

accruals regression as an additional variable. Due to the small sample size, the second

method is employed in this thesis.

Similar steps are involved. Firstly, it begins with the estimation of coefficients ά1, ά2,

ά3 and ά4 for each industry in each year by using OLS regression to extract the non-

discretionary accruals. Then, the error terms are estimated according to the difference

between the total accruals and the non-discretionary accruals, which represents the

discretionary component of the accruals. The model is as follows:

1 2 3 4 1

1 1 1 1

ijt ijt ijt ijt

ijt ijt

ijt ijt ijt ijt

TACC REV REC PPEά ά ά ά ROA e

TA TA TA TA

where:

DACC ij = discretionary accruals for sample firm i in industry j for year t-1;

ROAijt -1 = return on asset for sample firm i in industry j for year t-1;

ΔREVijt = change in revenues for sample firm i in industry j for year t;

year t;

4.5 Model specifications and related control variables

Three models of audit quality are used to test the relationship between the

effectiveness of the board of directors and the audit committee characteristics on audit

quality. These models are: the audit fees model, the NAS fees model and the auditor

industry specialist model. In order to examine the relationship between the board of

directors, the audit committee and auditor quality in constraining opportunistic

earnings, the earnings management model is employed..

A number of control variables are included in the models due to their real effect on

the estimation of audit quality and earnings management, all identified in the prior

literature. Since the prior studies differ in their use of control variables, in this thesis

the selection of the control variables are cited from key papers in the area of

investigation. If these variables were not controlled, it would likely lead to bias in

estimating audit quality and earnings management. Each of these models and related

control variables are described below.

4.5.1 Audit fees model

Simunic (1980) suggests that audit fees are a function of two basic components: (1)

the quantity of resources and (2) the expected future loss components. The first

component refers to the cost in audit resources associated with auditor effort or audit

hours and the second component refers to the projected likelihood of future losses that

the auditor may suffer, such as sanctions by regulatory agencies and litigation. These

two components can be categorised into three set of variables, namely size,

complexity and risk sharing (Simunic, 1980). DeFond et al. (2000) suggest that an

audit fees model that contains these variables has a higher explanatory power and is

more robust across samples, countries and time periods. Motivated by this reason,

below is the estimated models used to examine the relationship between the board and

audit committee characteristics, and audit fees:

LNAFEE = α0 + β1BRDSIZE + β2BRDNED + β3BRDEXP +

β4BRDMEET + β5ACSIZE + β6ACIND + β7ACEXP +

β8ACMEET + β9SQSUBS + β10FORGN + β11RECINV

+ β12LEVERG + β13LNASSET + ε

Dependent variable:

LNAFEE = the natural log of audit fees;

Hypothesis variables:

BRDSIZE = the numbers of board‟ members during the year;

BDRNED = the proportion of non-executive directors on board to board

BRDEXP = the proportion of directors with accounting experience and

financial qualification to board size;

BRDMEET = the number of board meetings during the year;

ACSIZE = the number of audit committee members;

ACIND = coded as 1 if audit committee had solely non-executive

directors; 0 otherwise;

ACEXP = the proportion of audit committee members with accounting

experience and financial qualification to audit committee

ACMEET = the number of audit committee meetings during the year;

Control variables:

SQSUBS = the square root of total consolidated subsidiaries;

FORGN = the proportion of foreign subsidiaries to total consolidated

subsidiaries;

RECINV = the proportion of total assets in account receivable and

inventory;

LEVERG = the proportion of debts to total assets;

LNASSET = the natural logarithm of total assets;

The choice of control variables captured the size, complexity and risk sharing factors

(Simunic, 1980). These variables are relatively similar to those adopted in Abbott and

Parker‟s (2003) study, which also examines audit fees and the characteristics of the

board and audit committee. However, the current study excludes the variable of audit

opinion, since all of the firms in the sample received an unqualified audit opinion.

Despite the fact that the variable of audit opinion is a proxy for risk sharing, the

present study replaces it with the LEVERG, consistent with Menon and Williams

(2001). The current study believes that LEVERG and RECINV variables are sufficient

to represent the risk sharing factor. All of the control variables are sourced from

Datastream except for the SQSUBS and FORGN, which are hand collected from the

annual reports.

The variables SQSUBS and FORGN are proxies for the complexity of firm operation.

SQSUBS is measured as the square root of the total consolidated subsidiaries and

FORGN is measured as the ratio of foreign subsidiaries to total subsidiaries. The prior

literature suggests that as the firms‟ business complexities increase, the auditors may

need to put more audit effort and audit hours into dealing with the complex business

operations, in turn resulting in higher audit fees (Simunic, 1980; Craswell and Francis,

1999; Abbott et al., 2003a; Carcello et al., 2002). These studies argue that the level of

audit effort increases with the geographical coverage and the number of transactions

in the subsidiaries. The present study expects these variables to be positively

associated with audit fees.

The variables RECINV and LEVERG are proxies for risk-sharing factors, where

RECINV is measured as the proportion of total assets in the account receivable and

inventory. According to Simunic (1980: 173), the account receivable and inventory

are the two “risky” items in the balance sheet that need more specific auditing

procedures. These items require the auditor to make confirmation, observation and the

valuation because it anticipates the forecast on the future events and may expose to a

higher material misstatement (Simunic, 1980). In addition to this, Pratt and Stice

(1994) claim that a higher RECINV is associated with a higher probability of audit

failure. Such considerable effort and the raised possibility of misstatement increase

the likelihood of audit failure, thus requiring more audit hours, resulting in increased

audit fees. In line with these arguments and the prior studies that control this variable

in their audit fees models (Simunic, 1980; Pratt and Stice, 1994; Palmrose, 1986a;

Francis and Stokes, 1986; Francis and Simon, 1987; Abbott et al., 2003a; Carcello et

al., 2002), the present study predicts a positive relationship between the variable

RECINV and audit fees.

LEVERGN is defined as the proportion of debts to total assets and measures the

clients‟ financial condition. According to Pratt and Stice (1991), clients in poor

financial condition can cause more frequent audit failure because the firms

experiencing the weak financial condition probably carry a higher risk of material

misstatement, which may be difficult for the auditors to detect. Prior literature treats

the leverage as the auditor‟ perceived litigation risk (Simunic and Stein, 1996; Menon

and Williams, 2001, Pratt and Stice, 1994). The higher the litigation risk, the higher

the auditors perceived they may involve in litigation. It is been argued that the

auditors increase the audit fees and audit effort in the firms that experience higher

leverage, in order to trade off the auditor‟ litigation risk. Consequently, the LEVERGN

is expected to be positively associated with the audit fees.

LNASSETS is a measure of firm size and is defined as the natural log of the total

assets (Simunic, 1980; Abbott et al., 2003a; Carcello et al., 2002). As the firm‟ size

increased, the auditors extend the audit scope and audit test accordingly. Such

extensive efforts increase audit hours and fees, and thus the present study predicts a

positive relationship between total assets and audit fees.

4.5.2 NAS fees model

Firth (1997) and Parkash and Venable (1993) suggest that NAS fees are the function

of agency cost, audit complexity, audit risk and demand for consulting services. In

line with these, the present study estimates the following model:

FEE = β0 + β1BRDSIZE + β2BRDNED + β3BRDEXP +

β4BRDMEET + β5ACSIZE + β6ACIND + β7ACEXP+

β8ACMEET + β9INOWN + β10BLOCK + β11LEVERG +

β12RETURN +β13LNASSET + β14 ACQ + β15NWFUND +

β16NEWDIR + β17RESTR + ε

Dependent variable:

FEE = LNNAF, LNTOTALFEES, FEERATIO1 and FEERATIO2.

BRDNED = the proportion of non-executive directors on board to board

BRDEXP = the proportion of directors with accounting experience and

financial qualification to board size;

ACEXP = the proportion of audit committee members with

accounting experience and financial qualification to audit

committee size;

Control variables:

INOWN = the cumulative percentage of total shares owned by the

directors of a firm;

BLOCK = the cumulative percentage shares ownership of the

blockholders who hold at least 5 percent or more of

outstanding common shares and who are unaffiiliated with

management;

RETURN = the fiscal year total stock return;

LNASSET = the natural logarithm of total assets;

ACQ = the number of acquisitions made by the company during

the year;

NWFUND = coded as 1 if the firm issued new shares or debt for cash

during the year, 0 otherwise;

NEWDIR = coded as 1 if the firm appoint new external director during

the year, 0 otherwise;

RESTR = coded as 1 if the restructuring program has occurred during

the year, 0 otherwise;

The choice of control variables is in line with the study conducted by Abbott et al.

(2003b), which examines the relationship between NAS and the characteristics of the

board and audit committee.43

These variables have been found by the prior literature to

be significant in explaining the magnitude of NAS purchased, which include INOWN,

BLOCK, LEVERG, RETURN, LNASSET, ACQ, ISSUE, NEWDIR and RESTR (Firth,

1997; Parkash and Venable, 1993). All of the control variables are obtained from

Datastream except for the variables ACQ and NEWISSUE, which are sourced

respectively from Thomson One Banker database and annual report.

The variables INOWN, BLOCK and LEVERG control for agency costs. INOWN is

measured by the cumulative percentage of total shares owned by the directors of a

firm at the beginning of the fiscal year. Jensen and Meckling (1976) suggest that the

higher the level of insider ownership, the lower the agency costs, since the directors of

externally owned firms will presumably realign their interests with the external

owners. Such alignment is likely to initiate greater monitoring because the directors

perceive themselves to be more accountable regarding their actions, and are thus

likely to demand a higher quality audit. Consistent with this argument, the present

study anticipates a negative relationship between INOWN and FEE.

However, the present study excludes the variable of the big-size auditor because all of the firms in

the sample are audited by big-size auditors.

BLOCK is defined as the cumulative percentage of ownership of the blockholders

who hold at least 5 % or more of outstanding common shares and are unaffiiliated

with management.44 Parkash and Venable (1993) claim that when outside ownership

is higher, the agency cost is likely to decrease because such significant ownership

provides incentives for direct monitoring. Greater monitoring appears to limit the

NAS purchased, as it seems that higher NAS compromises the independence of the

auditor, resulting in a low-quality audit. On the other hand, Abbott et al. (2003b)

claim that, as a result of information asymmetry between the management and outside

blockholders, the latter probably have limited access to insider information and are

therefore more likely to depend on the information provided by the management to

facilitate their monitoring function. Since there are two opposite arguments regarding

the BLOCK variable, the present study does not make any specific prediction about

the relationship between BLOCK and FEE.

LEVERG is also a measure for agency cost. Agency theory suggests that managers

have the incentive to transfer wealth from debtholders to shareholders using various

actions (Jensen and Meckling, 1976). As the amount of debt increases, the higher the

incentive of wealth will be transferred from the debtholders to shareholders and thus,

initiate greater demand for a highly independent auditor (Palmrose, 1986a; Simunic

and Stein, 1987; Francis and Wilson, 1988; DeFond, 1992). ). In other words, the

firms experiencing higher leverage are more likely to demand a higher quality auditor

or a more highly independent auditor in order to verify the accounting number related

to the covenant compliance. Since the prior literature suggests that the auditors‟

independence decreases as the amount of NAS purchased increases, the present study

predicts a negative relationship between LEVERG and FEE.

RETURN is a measure of firm performance and is defined by the total stock return for

the fiscal year. According to Houghton and Ikin (2001), the poor performance of firms

is related to ineffective and inefficient operations, inappropriate management

strategies, lack of competitiveness and the obsolescence of product and information

technology. These concerns might increase the incentive for firms to hire external

The blockholders are also known as „institutional investors‟ or large investors, rather than individual

investors, and include pension funds, mutual funds, corporations, insurance firms, private equity firms,

banks and trusts (Cronqvist and Fahlenbrach, 2009).

consultants for expert advice. Indeed, Firth (1997) argues that firms with a poor stock

market return are more likely to demand external advice on how to improve their

performance than firms with a higher stock return. Thus, the present study predicts a

positive relationship between RETURN and FEE.

LNASSET is a measure of firm size. The demand for NAS increases as the size of the

firm expands (Firths, 1997; Houghton and Ikin, 2001). When the size of the firm

grows, the firm becomes more complex and may thus demand greater NAS. FEE is

therefore expected to be positively related to LNASSET.

Corporate acquisition (ACQ), the issuance of new shares, the borrowing of cash

(ISSUE), the new appointment of external non-executive directors (NEWDIR) and

firm restructurings (RESTR) are the variables associated with the events that create the

demand for NAS (Firths, 1997). These variables represent relatively irregular events

but might potentially have a major impact on the firms. The present study expects

positive relationships between these variables and FEE.

4.5.3 Industry specialist model

Prior studies model the industry specialist auditor as a function of agency cost, audit

risk and firm‟ business complexities (Francis and Wilson, 1988; DeFond, 1992; Firth

and Smith, 1992). Following these studies, the model specification is as follows:

SPEC_AUD = β0 + β1BRDSIZE + β2BRDNED + β3BRDEXP +

β8ACMEET + β9INOWN + β10LNASSET + β11LEVERG +

β12NWFUNDRATIO + β13ROA + β14SQSUB +

β15FORNSALE + ε

Dependent variable:

SPEC_AUD = SPECLST_MSLEADER, SPECLST_30MS,

SPECLIST_MS, SPECLIST_PS and

SPECLST_WEIGHTED.

BRDNED = the proportion of non-executive directors on board to

board size;

BRDEXP = the proportion of directors with accounting experience

and financial qualification to board size;

accounting experience and financial qualification to audit

committee size;

Control variables:

INOWN = proportion of outstanding ordinary shares owned directly

by directors on the board;

LNASSET = natural logarithm of total assets;

LEVERG = proportion of debts to total assets;

NWFUNDRATIO = proceed from new debt and equity issuances/ total assets;

ROA = return on assets;

SQSUB = square root of total consolidated subsidiaries;

FORGNSALE = proportion of the firm‟ foreign sales.

The control variables chosen are relatively similar to those chosen by Abbott and

Parker (2000) and Chen et al. (2005), and include INOWN, LNASSET, LEVERG,

NWFUND, ROA, SQSUB and FORGNSALE. The theoretical justifications of these

variables stem from previous related studies on big-size auditors (Francis and Wilson,

1988; DeFond, 1992; Firth and Smith, 1992).45

INOWN, LNASSETS and LEVERG are the proxies for the agency variables. INOWN is

a measure for the insider ownership, and there are two arguments to explain the

relationship between INOWN and SPEC_AUD. Firstly, as the insider ownership

increases, the directors align their interests with those of the external owners, and thus 45

As explained in Chapter 3 (see page 66), the theoretical foundation for industry specialist auditors is

derived from reputation capital theory among the big-size auditors.

initiate greater monitoring of management actions (Jensen and Meckling, 1976). Since

they perceive themselves as part of the firm structure, they demand a higher quality

audit, engaging the industry specialist auditor. This indicates a positive relationship

between INOWN and SPEC_AUD (Abbott and Parker, 2000). The second argument

suggests that as the insider ownership increases, the directors obtain more detailed

inside information about the firms (Firth, 1997), and hence have less need for a higher

quality auditor. This argument leads to a negative relationship between INOWN and

SPEC_AUD. Since there are two possible arguments, no prediction is made for this

variable.

LNASSET is a proxy for firm size. Chen et al. (2005) argue that larger firms are likely

to engage a higher quality auditor because the agency costs increase as the firm size

grows (Francis and Wilson, 1988; Firth and Smith, 1992). The higher the agency cost,

the higher the demand for industry specialist auditors. The present study therefore

predicts a positive relationship between LNASSET and SPEC_AUD.

LEVERG is the ratio of total debt to total assets. As the level of leverage increases, the

agency costs also increase as a result of the increased likelihood of wealth transferral

from debtholders to shareholders. Such conditions create a demand for a higher

quality audit to verify the accounting numbers in the debt contract by reducing the

information asymmetry between the debtholders and managers. In line with Abbott

and Parker (2000) and Chen et al. (2005), the present study controls LEVERG for the

extent of agency cost in the auditor industry specialist model. The present study

expects that the leverage will be positively related to the engagement of industry

specialist auditors.

NWFUND is defined as the firm‟s acquisition of new debts and equity issuances, in

proportion to the firm‟s total assets. Titman and Trueman (1986) claim that the

appointment of a higher quality auditor signal credibility of information to the users.

Thus, several studies suggest that by engaging a higher quality auditor, i.e. an industry

specialist auditor, the firms may increase the marketability of new securities to

outsiders, because these firms are perceived to have more credible information

(DeFond, 1992; Francis and Wilson, 1988; Johnson and Lys, 1990). It is been argue

that the monitoring function by the reputable auditor increased the credibility of the

firm‟ information and thus more favour in making new securities or obtaining new

borrowing. Consistent with Abbott and Parker (2000) and Chen et al. (2005), the

present study predicts a positive relationship between NWFUND and SPEC_AUD.

ROA is a measure of client profitability and risk. According to Abbott and Parker

(2000), more profitable firms are more likely to engage industry specialist auditors

because they are more willing to pay the fee premium. On the other hand, ROA could

also act as a risk-sharing proxy. Some studies suggest that riskier firms are more

likely to appoint a higher quality auditor in order to signal the credibility of

information to outsiders (Datar et al., 1991; Hogon, 1997; Copley and Douthett,

2002). These studies argue that the demand for a higher quality auditors increases

with the firm risk. Since there are two possible arguments, the direction of this

variable is not predicted.

Similar to the audit fees model, SQSUB and FORGNSALE are the proxies for firm

complexity. The more complex the firms‟ operation, the greater the need for a higher

quality auditor (Collier and Gregory, 1996; Simon and Francis, 1988). Following

Abbott and Parker (2000) and Chen et al (2005), it is been argue that the

diversification of the clients‟ subsidiaries and higher foreign sales are expected to

increase the complexity of the firms‟ financial reporting system, the audit program

and audit scope, and these create the greater need for the industry specialist auditor.

The present study expects a positive relationship between these two variables and

engagement of industry specialist auditor.

4.5.4 Earnings management model

The earnings management model follows that of Klein (2002) and Bédard et al.

(2004), who examine the effects of the board and audit committee characteristics and

the auditor quality on opportunistic earnings. The model specification is as follows:

DACC = β0 + β1BRDSIZE + β2BRDNED + β3BRDEXP +

β8ACMEET + β9AQ + β10INOWN + β11BLOCK + β11MTBV+

β13LOSS + β14CFO + β15LEVERGN + β16LNASSET+ ε

Dependent variable:

DACC = DACCJM, DACCMJM and DACCROA

AQ = LNAFEE, LNNAF, LNTOTALFEES, FEERATIO1,

FEERATIO2, SPECLST_MSLEADER,

SPECLST_MS30, SPECLST_MS, SPECLST_PS and

SPECLST_WEIGHTED

BRDNED = the proportion of non-executive directors on board to

board size;

BRDEXP = the proportion of directors with accounting experience

and financial qualification to board size;

accounting experience and financial qualification to

audit committee size;

ACMEET = the number of audit committee meetings during the

Control variables:

INOWN = the cumulative percentage of total shares owned by the

directors of a firm;

BLOCK = the cumulative percentage shares ownership of the

blockholders who hold at least 5 percent or more of

outstanding common shares and who are unaffiiliated

with management;

MTBV = the market to book value ratio;

LOSS = coded as 1 if the firm had two or more years of negative

income, 0 otherwise;

CFO = cash flow from operation scaled by lagged total asset;

LNASSET = natural logarithm of total assets;

The control variables chosen are relatively similar to those used by Klein (2002) and

Bédard et al. (2004), and include INOWN, BLOCK, MTBV, LOSS, CFO, LEVERG

and LNASSET. Most of the prior studies on earnings manipulation find these variables

to be significantly correlated with the level of discretionary accruals (e.g. Ferguson et

al., 2004; Park and Shin, 2004).

Similar to the NAS fees and industry specialist auditor models, INOWN and BLOCK

are the variables grounded from the agency theory. As the level of insider or

managerial ownership increases, they are more closely aligned with the interest of the

external shareholders and thus, less likely to pursue opportunistic behaviour at the

expenses of the shareholders (Jensen and Meckling, 1976). Evidence suggests that

managers with a higher level of ownership are more likely to increase the earnings

informativeness and report more reliable earnings (Warfield et al., 1995). This

situation arise because they perceive themselves as part of the firm‟s structure and

therefore more responsible for their actions. In agreement with Klein (2002), the

present study predicts a negative relationship between INOWN and earnings

management.

From the perspective of agency theory, the institutional investors or BLOCK could be

an alternative monitoring incentive. According to Monks and Minow (1995), the

institutional investors have the opportunity, resources, ability to monitor, discipline

and influence management actions and decisions. When outside ownership is

concentrated, such significant and collective shareholdings provide institutional

investors with a greater incentive to collect information, to monitor activities and to

encourage better performance from the management (Chung et al., 2002). Stated

differently, as the cumulative shareholdings of the institutional investors increases, the

more likely they are to constrain the management opportunistic behaviour, which in

turns resulting lower earnings management. Consistent with this argument and

following Klein (2002) and Bédard et al (2004), BLOCK is expected to be negatively

correlated with the levels of opportunistic earnings.

MTBV is defined as the market-to-book value ratio and used as a proxy for growth

opportunities. Skinner and Sloan (2002) and Matsumoto (2002) suggest that the

managers of firms with higher growth opportunities face greater pressure to meet

earnings targets. Thus, the higher the market-to-book value of equity ratio, the greater

the incentive for managers to manipulate earnings. In line with Klein (2002), Collin

and Kothari (1989) and Graver and Graver (1993), the present study expects that

MTBV will be positively related to earnings management.

LOSS is a dichotomous variable, coded as 1 if the firm had two or more years of

negative income, 0 otherwise. The firms with negative income may have greater

incentive to manage reported earnings in order to survive in the market. Burgstahler

and Dichev (1997: 124) claim that 30% to 40% of firms with slightly negative

earnings exercise discretion to report positive earning. If the firms report negative

earnings they may unable to grant loans and the investors are unfovour towards the

negative earnings. This may suggest that the firms with negative income will engage

more in earnings management as compared to the firms with positive income, in order

to grant loans and promote the shareholders interests. Thus, the current study expects

LOSS to be positively associates to DACC.

CFO is defines as cash flow from the firm‟soperation scaled by lagged total assets.

The managers with lower cash flow have a higher incentive to manipulate earnings by

reporting future revenues or by delaying current costs in order to report they are in

well financial condition (Leuz et al., 2003). This argument suggests a negative

relationship between CFO and DACC, consistent with the evidence documented in

Becker et al. (1998). Alternatively, the firms with a higher cash flow may also

manipulate earnings or unstate the strong performance by creating a reserve for their

future needs (Leuz et al., 2003). This is supported by Han and Wang‟s study (1998).

They claim that the petroleum refining firms that benefited from higher oil prices used

income-decreasing accruals to mitigate the potential of political risk. In addition,

Frankel et al. (2002) argue that firms with a high cash flow are more likely to beat the

earnings benchmark. This may suggest a positive relationship between firm

performance and earnings management, but the mixed arguments offer no direction

for this variable

LEVERG is used as a proxy for debt covenants violation. Press and Weintrop (1990)

and Duke and Hunt (1990) claim that a firm with higher leverage is more likely to be

associated with the debt covenants violation. This is because as the level of debt

increases, the firm may experience tighter accounting constraints, which in turn

increases the higher possibility of debt covenant violations. Several studies suggest

that in order to avoid violating restrictive debt covenants, the higher leveraged firms

are more likely to choose accounting procedures that support income increasing

accounting method (Bowen et al., 1981; Dhaliwal et al., 1982). In particular, DeFond

and Jiambalvo (1994) suggest that the highly leveraged firms have a greater incentive

to make income increasing discretionary accruals in order to avoid debt covenants

violation. This may suggest a positive relationship between leverage and earnings

management. However, DeAngelo et al. (1994) claim that in a time of financial

distress, firms might engage in contractual renegotiations with lenders in order to

deliberately reduce reported earnings. Such a situation may suggest a negative

relationship between leverage and earnings management. Park and Shin (2004) report

a negative relationship between leverage and earnings management for different

reasons. They argue that highly leveraged firms might be less likely to manage their

earnings because they are under the close scrutiny of their lenders. If the lender has

closely monitored the earnings management activities of such a firm, then their

earnings management will decrease with financial leverage. As a result of such

differing arguments, the direction of this variable is not predicted for the current

study.

LNASSET is a proxy for firm size. The larger the firm size, the higher the likelihood

that the manager will manipulate the firm‟s earnings .Watt and Zimmerman (1990)

suggest that larger firms are associated with higher political costs, and that there is

thus a higher incentive to manipulate reported earnings in order to mitigate potentially

adverse political actions. Evidence from other prior studies also suggests a positive

relationship between size and earnings management (Becker et al., 1998; DeFond and

Park, 1997). However, Park and Shin (2004) provide an alternative argument. They

claim that larger firms are followed by the external capital market, and are therefore

less able to hide earnings manipulation because they are closely monitored by the

press and by analysts. This suggests a negative relationship between LNASSET and

DACC. The mixed arguments propose no clear direction regarding the association

between LNASSET and DACC.

4.6 Data analysis procedures

To analyse the data, the statistical software STATA 11 is used. The data analysis

includes descriptive statistics, the correlation matrix, multivariate regression and

robustness tests. Each of these is now reviewed.

4.6.1 Descriptive statistics and correlation matrix

Descriptive statistics describes the sample data on a single variable in an organised

form. It includes the mean, median, standard deviation, minimum, maximum,

skewness, kurtosis, first quartile and third quartile. The mean, median, first quartile,

third quartile and standard deviation measure the central tendency of the variable. The

skewness and kurtosis explain the shape of the data distribution. Specifically, the

skewness measures the symmetry of distribution while kurtosis measures the

peakedness or flatness of the distribution (height), as compared to normal distribution

(Hair et al., 2010:35-36). In addition to descriptive statistics, when the dependent

variables are dichotomous and can be regarded as „low‟ or „high‟ groups, univariate

analyses will be performed to test the mean differences between these two groups.

The correlation among the variables is shown by pairwise correlation matrix. This

explains the degree of linear association between two variables and ranges from +1 to

-1, where a correlation of ±1 means that there is a perfect linear relationship between

the variables. However, according to Hair et al. (2010: 200), a higher degree of

intercorrelation among the independent variables may cause problems of

multicollinearity when the correlation coefficient is above ±0.90. Multicolinearity

may substantially effect the predictive ability of the regression model as well as the

estimation of the regression coefficients.

4.6.2 Multivariate regression

Most of the multivariate regression in the prior literature used ordinary least square

regression (hereafter OLS) to examine the relationship between a single dependent

variable and several independent variables (predictors). However, there are five

fundamental assumptions to be fulfilled in order for the OLS regression model to be

valid (Chen et al., 2003; Gujarati, 2003; Hair et al., 2010). These assumptions include:

(1) Normality - The errors (residuals) should be normally distributed

(2) Linearity - The relationship between the predictors and the response variable

should be linear.

(3) Homoscedasticity - The error variance should be constant

(4) Independent - The errors associated with one observation should not be

correlated with the errors of other observations.

(5) Multicollinearity - There should be no exact collinearity among predictors.

In addition to these assumptions, Chen et al. (2006) suggest that the analyst consider

any unusual and influential data that may substantially change the estimation of

coefficients. When these assumptions are violated, the results of OLS regression may

be distorted and biased (Chen et al., 2003; Gujarati, 2003; Hair et al., 2010).

Several regression estimators, such as least square regression with robust standard

error, weighted least square regression (sometimes known as generalized least square

regression or GLS), robust regression, and quantile regression provide an alternative

to OLS regression when the assumptions have been violated. For example, when the

normality assumption has not been fulfilled, and in the presence of moderate outliers,

robust regression (iteratively reweighted least squares) provides better estimation than

OLS regression (Hamilton, 1999; Chen et al., 2003). In the presence of

heteroscedasticity and serial correlation, either the least square estimator with robust

standard error (Huber-White standard errors) or GLS regressions are able to reweight

the error variance and thus to correct heteroscedasticity and autocorrelation (Adkins

and Hill, 2007: 196; Gujarati, 2003: 387). In addition to these, non-parametric

regression, such as quantile regression disregards all the OLS assumptions (Gujarati,

2003).

In general, the present study finds that most of the OLS assumptions are not

adequately fulfilled, even though several steps have been taken to conform to these

assumptions (by using data transformation, for example). In the prior literature, many

statisticians agree that mild violations of the OLS assumptions are robust and

unaffected in many situations (Box, 1953; 1954; Glass and Hopkins, 1984; Glass et

al., 1972; Newman et al., 1989). Thus, in the main analysis of this study, most models

have been analysed using OLS regression, except the heteroscedastic models, where

the use of OLS regression would be questionable. Where the models reveal

pronounced heteroscedasticity, the analysis will be performed using the least square

estimator with robust standard error. This regression estimates the standard errors

using the Huber-White sandwich. It can be interpreted in the same manner as OLS

regression, but is more precise and efficient than the OLS estimator and able to

correct heteroscedasticity and autocorrelation (Adkins and Hill, 2007). In the

sensitivity analysis, the alternative regression estimators such will be reported as a

benchmark for comparison.

OLS regression is applicable when the dependent variable is continuous variable.

However, when the dependent variable is a dichotomous variable, OLS regression

may not be able to fulfil the OLS assumptions and may thus lead to inefficient

estimation (Pampel, 2000; Menard, 2002). Thus, transforming the dichotomous

variable into a logit or probit model may overcome the inefficiency. As pointed out by

Menard (2002: V):

“… with a dichotomous dependent variable, assumptions of homoskedasticity,

linearity, and normality are violated, and OLS estimates are inefficient at best.

The maximum likelihood estimation of a logistic regression overcomes this

inefficiency, transforming Y (1, 0) into a logit (log of the odds of falling into the

„1‟ category).”

Therefore, taking into account these circumstances, when the dependent variable is

dichotomous, multivariate regression is estimated using the heteroskedastic ordinal

regression as a control for heteroscedasticity (Williams, 2009). In the sensitivity

analysis, probit regression is used as an alternative estimator. According to Pampel

(2000: 54) logit and probit regression give essentially similar results and it is up to the

researcher to choose between these two estimators.

4.6.3 Further analysis and robustness test

Several tests were performed after the multivariate regression analysis. The purpose

of these additional tests was to give reasonable assurance that the main findings were

robust to the various model specifications. The robustness tests include tests for

multicollinearity and heteroscedasticity, various regression estimators, client size

analysis, various definitions of corporate governance characteristics, and tests for

additional control variables and endogeneity.

4.7 Summary

The sample population for this thesis is the largest 350 firms listed on the LSE, giving

an initial sample of 1,400 firm-year observations over the period 2005 to 2008. After

eliminating regulated firms, firms with missing information and firms not audited by

the Big 4 auditors, the final sample consisted of 674 and 613 firm-year observations

for two empirical investigations. The data was gathered from Datastream, FAME,

Thomson One Banker and the firms‟ annual reports. The computation and description

of three main groups of hypothesis variables have been explained in this chapter: (1)

the characteristics of the board of directors and the audit committee (e.g. size,

composition, expertise and level of activities), (2) the audit quality proxies (e.g. audit

fees, NAS fees and industry specialist auditors), and (3) earnings management. This

has been followed by a discussion of the model specifications and related control

variables. Table 4.6 presents a summary of the variables used in the current study.

Most of the analyses employed OLS regression and the least square estimator with

robust standard error (Huber-White sandwich).

Table4.6: Summary of all variables used

Variable Label Description Data source

Board of

directors size

BRDSIZE the numbers of board‟ members

during the year;

Annual

report

Board of

director

composition

BDRNED the proportion of non-executive

directors on board to board size;

Annual

report

Board of

directors

expertise

BRDEXP the proportion of directors with

accounting experience and financial

qualification to board size;

Annual

report

Board of

directors

meeting

BRDMEET the number of board meetings during

the year;

Annual

report

committee

ACSIZE the number of audit committee

members;

Annual

report

committee

composition

ACIND coded as 1 if audit committee had

solely non-executive directors; 0

otherwise;

Annual

report

Table4.6 (continued)

committee

expertise

ACEXP the proportion of audit committee

members with accounting experience

and financial qualification to audit

committee size;

Annual

report

committee

meeting

ACMEET the number of audit committee

meetings during the year;

Annual

report

Audit fees LNAFEE the natural log of audit fees FAME

NAS fees LNNAF natural log of NAS fees; FAME

NAS fees LNTOTALFEES natural log of the sum of audit and

NAS fees

NAS fees FEERATIO1 the fee ratio of NAS fees to total fees FAME

NAS fees FEERATIO2 the fee ratio of NAS fees to audit fees FAME

Industry

specialist

auditors

SPECLST_

MSLEADER

coded as 1 if the auditor earned the

largest market share in each particular

industry, 0 if otherwise;

Industry

specialist

auditors

SPECLST_MS30 coded as 1 if the auditor‟ market share

exceed 30 percent in each particular

industry, 0 if otherwise;

Industry

specialist

auditors

SPECLST_MS continuous variable which equals to

the respective auditor‟ market share;

Industry

specialist

auditors

SPECLST_PS continuous variable which equals to

the respective auditor‟ portfolio share

Industry

specialist

auditors

SPECLST_

WEIGHTED

continuous variable which equals to

the compliment between auditor‟

market share (SPECLST_MS) and

portfolio share (SPECLST_PS)

Earnings

management

DACCJM discretionary accrual based on Jones

Model;

Datastream

Earnings

management

DACCMJM discretionary accruals based on

Modified Jones model;

Datastream

Earnings

management

DACCROA discretionary accruals by Kothari et al.

(2005), including lagged ROA in the

accrual regression to control for firm

performance;

Datastream

Consolidated

subsidiaries

SQSUBS the square root of total consolidated

subsidiaries;

Datastream

Foreign

subsidiaries

FORGN the proportion of foreign subsidiaries

to total consolidated subsidiaries;

Datastream

Ratio of

account

receivable

inventory to

total asset

RECINV the proportion of total assets in

account receivable and inventory;

Datastream

Table4.6 (continued)

Leverage LEVERG the proportion of debts to total

assets;

Datastream

Total asset LNASSET the natural logarithm of total assets; Datastream

Board of

director

ownership

INOWN the cumulative percentage of total

shares owned by the directors of a

Annual

report

Blockholder

ownership

BLOCK the cumulative percentage shares

ownership of the blockholders who

hold at least 5 percent or more of

outstanding common shares and who

are unaffiiliated with management;

Annual

report

Stock return RETURN the fiscal year total stock return; Datastream

Number of

acquisition

ACQ the number of acquisitions made by

the company during the year;

Thomson

One Banker

New shares

or debt for

NWFUND coded as 1 if the firm issued new

shares or debt for cash during the

year, 0 otherwise;

Annual

report

New external

director

NEWDIR coded as 1 if the firm appoint new

external director during the year, 0

otherwise;

Annual

report

Restructuring RESTR coded as 1 if the restructuring

program has occurred during the

year, 0 otherwise;

Annual

report

New debt

and equity

NWFUNDRATIO proceed from new debt and equity

issuances/ total assets;

Annual

report and

Datastream

Return on

assets

ROA return on assets; Datastream

Foreign sale FORGNSALE the proportion of the firm‟ foreign

sales.

Datastream

Growth MTBV the market to book value ratio; Datastream

performance

LOSS coded as 1 if the firm had two or

more years of negative income, 0

otherwise;

Datastream

Cash flow CFO cash flow from operation scaled by

lagged total asset;

Datastream

Liquidity LIQ ratio of current assets divided by

current liabilities;

Datastream

Sales growth GROWTH growth rate in sales over the

previous fiscal year;

Datastream

CHAPTER 5

FINDINGS AND DISCUSSIONS: THE BOARD OF DIRECTORS, AUDIT

COMMITTEE AND AUDIT QUALITY

5.1 Introduction

This chapter presents the results for the first empirical analysis of the characteristics

of boards of directors and audit committees and their relationship to audit quality.

There are three proxies of audit quality to be examined, namely the audit fees, the

NAS fees and the use of industry specialist auditors.

This chapter is organised as follows: the next section presents the descriptive statistics

and correlation matrix. This is followed by a separate section on research design,

multivariate results and a sensitivity analysis of each proxy. The last section

summarises and concludes the chapter.

5.2 Descriptive statistics

Table 5.1 present the descriptive statistics of three measures of audit quality- audit

fees (LNAFEE), NAS or auditor independence measures (LNNAF, LNTOTALFEES,

FEERATIO1, FEERATIO2) and auditor industry specialist measures

(SPECLST_MSLEADER, SPECLST_MS30, SPECLST_MS, SPECLST_PS,

SPECLST_WEIGHTED, the hypothesis variables (BRDSIZE, BRDNED, BRDEXP,

BRDMEET, ACSIZE, ACIND, ACEXP, ACMEET) and the related control variables

containing mean, median, standard deviation, minimum and maximum, skewness and

kurtosis. The following are important descriptive statistics to highlight.

The mean (median) of audit fees and NAS fees for 674 firm-years are £1.466 million

(£0.805 million) and £1.296 million (£0.600 million) respectively. The mean (median)

of total fees is £2.762 million (£1.521 million). The NAS fees captured almost 50% of

total audit fees. In a previous UK study, O‟ Sullivan (2000) reported that the mean

(median) of audit fees and NAS fees for the 402 largest firms in the fiscal year of

1992 are £0.638 million (£0.279 million) and £0.320 million (£0.144 million)

respectively. As compared to O‟Sullivan (2000), audit fees have increased about

130% and the NAS fees of UK firms have grown by about 305%, suggesting the

importance of NAS as an alternative source of income for auditors in addition to

auditing services. These figures support the argument that NAS seems to be

associated with luxury income as it contains a higher profit margin than audit fees

(Lee, 2008). The mean (median) ratio of NAS fees to total fees (FEERATIO1) and

ratio of NAS fees to total audit fees (FEERATIO2) are 0.418 (0.400) and 1.0810

(0.667) respectively. Under the SPECLST_MSLEADER and SPECLST_MS30

definitions, 40.6% and 50% of firms, respectively, have engaged the services of an

industry specialist auditor.

For board of director variables, the average board size is 9, which is relatively

consistent with the figure reported in Peasnell et al. (2005) who report that the average

board size was 8. The average proportion of independent non-executive directors on

boards is 46%.46

The average proportion of boards of directors with accounting or

financial qualifications and experience is 35% and the average frequency of board

meetings is 8 times a year. This can be compared to the US studies of Abbott et al.

(2003) and Carcello et al. (2002) who report that the proportions of independent non-

executive directors on boards are 68% and 75% respectively, and that on average

board meetings are held 7 times a year.47

This comparison implies that firms in the US

are more likely to be dominated by independent non-executive directors, while in the

UK, board members have an almost balanced representation of executive and

independent non-executive directors. The board meeting frequencies are relatively

similar.

With regard to audit committee‟s variables, the mean (median) audit committee size is

3.635 (3.000). 72.2% of the samples report that their audit committees are made up of

solely independent non-executive directors with an average of 39% of them having

accounting and financial expertise; or at least 93% of the samples had one member

with financial expertise. The average frequency of audit committee meetings is 4

Using the 1992 data from 402 of the largest UK firms, O‟Sullivan (2000) reports that the percentage

of non-executive directors on boards was 41%. Similarly, Peasnell et al (2005), report that the

percentage of non-executive director on boards in their sample was 43%. However, both studies do not

mention whether they include the grey or affiliated directors.

47 Abbott et al. (2003) and Carcello et al. (2002) define the independent non-executive directors as non-

employee directors while the present study measures the independence of non-executive directors more

specifically using the definition outlined in the UK Corporate Governance Code (2010).

times a year. In US, Abbott et al. (2003) report that 75% of their samples had audit

committees that were composed of solely independent non-executive directors and

80% had at least one financial expert.48

This suggests that the proportion of firms with

a percentage of solely independent audit committee members are relatively similar in

the US and the UK but that the proportion of firms that have at least one audit

committee equipped with financial expertise in the UK are relatively higher than is

reported in the US. One of the possible reasons for this is because the sample

represented in this study comprised current data where the board and audit committee

system in the UK is already established, following the recommendations set out in the

UK Combined Code in July 2003.

The mean (median) of the cumulative percentage of total shares owned by the director

(INOWN) and the blockholders (BLOCK) are 4.201% (0.230%) and 23.159% and

(20.64%) respectively. In an earlier UK study, O‟Sullivan (2000) reported that the

mean (median) of shares owned by the directors was 5.55% (0.291%), while the

figures for shares owned by external shareholders were about 31.292% (21.640%).49

The mean (median) of INOWN is relatively similar but the mean (median) of BLOCK

is relatively lower than reported by O‟Sullivan (2000). The inconsistent mean

(median) may be due to different definitions of variables. The present study defines

BLOCK as the blockholders holding at least at 5% of the shares and O‟Sullivan

(2000) defines this at 3% of the shares. Allowing for the time difference from the

previous UK study, it shows that there was not much change in the directors‟

ownership pattern.

The mean (median) of total assets is 4,337 (1,379) millions. In natural logarithm form,

the mean (median) of LNASSET is 6.174 (6.129) of which 29.7% are represented by

the account receivable and inventories. The mean (median) of FORGNSALE is 45.631

(50.205), which captured nearly half of the total sales made by the 674 firms.

Abbott et al (2003) examined data 2001.

49 O‟Sullivan split the directors‟ shares into executive and nonexecutive directors and the block

shareholders into institutional and external shareholders. To make a comparison the present study adds

the mean and median of the respective variables.

The mean (median) for leverage is 64% (64.2), which is relatively higher than that

reported in Australia. Chen et al. (2005) report that the mean (median) leverage of 458

Australian firms is 48.84% (44.42%). The higher leverage figures may suggest that

the firms in the UK have higher risk levels than firms in Australia. In terms of firm‟s

performance, the mean (median) of ROA is 9.744 (8.34).

The mean (median) of fiscal year stock returns is 12.853 (7.477). For corporate

acquisition (ACQ) the mean (median) is 1.195 (1.000), approximately and on average

every firm will acquire one firm. Previously, Fifth (1997) reported that, for the 500

largest British industrial firms in 1993 (listed firms ranked in The Times 1000), the

means of RETURN and ACQ are 0.43 and 0.13 respectively. While Abbott et al.

(2003b) reported that the means of these variables are 0.0415 and 0.2402 respectively.

Compared to prior studies, the means of RETURN and ACQ that are documented in

the present study are relatively higher. This difference may be due the sample

variations. The firms listed in The Times 1000 and the sample examined by Abbott et

al (2003b) probably contained both the largest and the smaller firms. The present

study examined the largest 350 firms listed in LSE, thus, on average, the largest firms

may have a more stable stock market return and a higher number of acquisitions than

smaller firms.

NEWFUND, NEWDIR and RESTR are dichotomous variables. These variables are

associated with special events that demand NAS. The means of these variables are

0.922, 0.472 and 0.409 respectively. Specifically, there are 642 firm-years in which

new shares or debt for cash was issued, 318 firm-years in which new directors were

appointed and 276 firm-years in which a restructuring program took place. Fifth

(1997) and Abbott et al. (2003b) reported that the means of these variables are

0.11and 0.0851for ISSUE, 0.09 and 0.0408 for NEWDIR, and 0.05 and 0.0887 for

RESTR respectively.

The mean and median of the total number of consolidated subsidiaries are 26.866 and

20 respectively, and about 52.2% of them are foreign subsidiaries. The previous UK

and Australian studies, O‟Sullivan (2000) and Chen et al (2005) report that the overall

numbers of subsidiaries in their samples are 23.686 and 27.72 respectively. This

suggests that the levels of firms‟ complexities in the UK and Australia are relatively

similar.

As shown in the skewness and kurtosis columns in Table 5.2, most of the variables are

not normally distributed. The normal distribution takes a value of 0, and values above

and below 0 denote departures from normality. To reach the normal distribution,

several variables are been transformed (e.g. LNAFEE, SQSUBS, LNASSET) using the

natural log and the square root. One possibility that violates the normality assumption

may cause the heteroscedasticity problem. Diagnostics on the heteroscedasticity

problem will be provided in a later section.

5.3 Correlation matrix

The correlation matrix for all the variables used in the audit quality models is

presented in Table 5.2. The hypotheses variables and related control variables are

measured by three proxies of audit quality, namely audit fees (LNAFEE), auditor

independence measurements (LNNAF, LNTOTALFEES, FEERATIO1 and

FEERATIO2) and auditor industry specialist measurements (SPECLST_MSLEADER,

SPECLST_MS30, SPECLST_MS, SPECLST_PS and SPECLST_WEIGHTED). Higher

correlations among audit quality measurements are always expected since they are

highly interrelated. Only one measure of audit quality will be included in a single

empirical test so that this higher correlation does not necessarily influence the

empirical results.

In general, the overall correlation matrix shows that each of the audit quality measures

with all variables (i.e. board of directors, audit committee and related control

variables) are moderately inter-correlated with one and another except for variables

LNAFEE and LNTOTALFEES with LNASSET (correlation coefficients of 74.6% and

72.5% respectively) and FORGN with FORGNSALE (correlation coefficients of

71.3%), which have the largest correlation coefficients above 50%.50

The main

concern arising from the largest correlations is that they may indicate a

multicollinearity problem in the empirical model. Diagnostics on the multicollinearity

The highest correlation coefficient between FORGN and FORGNSALE is not critical because these

variables are associated with different model specifications. FORGN is one of the control variables for

the audit fees model while FORGNSALE is a control variable for the auditor industry specialist model.

that is associated with each empirical model will be provided later in the section on

additional analyses.

LNAFEE is significantly correlated with all board and audit committee variables

except for BRDMEET. The range of the correlation coefficient is between -8.9% and

46.7%. LNAFEE is significantly and positively correlated with BRDSIZE and

BRDNED but negatively correlated with BRDEXP. A positive correlation between

BRDNED and LNAFEE is consistent with Abbott et al. (2003a) and Carcello et al.‟s

(2002) studies. LNAFEE is significantly and positively correlated with ACSIZE,

ACIND and ACMEET (correlation coefficients of 41.4%, 10.6% and 33.7%

respectively), suggesting that the firms with a larger audit committee made up solely

of independent members and having more meetings are correlated with higher audit

fees. These correlations are consistent with the findings of Abbott et al. (2003a).51

However, the ACEXP is found to be negatively correlated with LNAFEE (at p<0.05),

and this is consistent with the correlation reported by Krishnan and Visvanathan

(2009).

LNNAF and LNTOTALFEES are found to be significantly correlated with most of the

board and audit committee variables when compared to FEERATIO1 and

FEERATIO2. LNNAF and LNTOTALFEES are positively correlated at p<0.01 with

BRDSIZE and BRDNED. Their correlation coefficients are 31.0% and 46.6%

respectively. A positive correlation between LNNAF and BRDNED is consistent with

O‟Sullivan (2000). BRDEXP is significantly and negatively correlated with LNNAF

and LNTOTALFEES (correlation coefficients are -10.6% and -16.7% respectively),

indicating that the firms with a higher proportion of board members with financial

expertise are likely to report lower NAS fees and total fees. BRDMEET is

insignificantly correlated with LNNAF and LNTOTALFEES. None of the board of

director variables is significantly correlated with FEERATIO1 and FEERATIO2.

ACSIZE and ACIND are significantly and positively correlated with LNNAF and

LNTOTALFEES but ACSIZE is found to be negatively correlated with FEERATIO2

Except for audit committee size, Abbott et al. (2003a) was not examined this variable.

The correlation coefficients of the board and audit committee variables with the

auditor industry specialist measures are mixed, depending on how the auditor industry

specialist data has been computed. They either marginally correlated in the opposite

directions or insignificantly correlated with board and audit committee variables. For

example, BRDSIZE is found to be significantly and positively correlated with all the

auditor industry specialist measures except for SPECLST_PS, which is found to be

insignificantly correlated with BRDSIZE. BRDEXP is negatively correlated with

SPECLST_MS, SPECLST_MSLEADER and SPECLST_MS30 (correlation coefficients

are -10.0%, -14.4%, -7.8% respectively) but it is positively correlated with

SPECLST_PS (the correlation coefficient is 7.8%). BRDNED is insignificantly

correlated with all the auditor industry specialist measures. For audit committee

variables, ACSIZE is significantly and negatively correlated with SPECLST_PS but it

is positively correlated with SPECLST_PS, SPECLST_MSLEADER and

SPECLST_MS30. ACIND is found to be positively and significantly correlated only

with SPECLST_PS, SPECLST_WEIGHTED, and SPECLST_MS30. ACEXP and

ACMEET are positively correlated with SPECLST_PS.

Table 5.1: Descriptive statistics (N=674)

Variables Mean Median Standard

Deviation Minimum Maximum Skewness Kurtosis

Audit fees (£‟000) 1466.024 805.500 1920.880 70.000 17000.000 3.320 18.201

NAS fees (£‟000) 1296.004 600.000 2058.770 0.000 25000.000 4.579 37.131

Total fees (£‟000) 2762.028 1521.000 3631.154 117.000 30000.000 3.279 17.411

LNAFEE 2.920 2.906 0.456 1.845 4.230 0.203 2.606

LNNAF 2.699 2.778 0.729 0.000 4.397 -1.309 6.406

LNTOTALFEES 3.189 3.182 0.466 2.068 4.477 0.137 2.635

FGEERATIO1 0.418 0.400 0.199 0.000 0.917 0.212 2.589

FEERATIO2 1.080 0.667 1.307 0.000 11.113 3.661 21.792

SPECLST_MS 0.349 0.321 0.215 0.015 0.903 0.704 2.835

SPECLST_PS 0.170 0.133 0.147 0.003 0.537 1.037 3.239

SPECLST_WEIGHTED 0.064 0.045 0.064 0.000 0.228 1.172 3.281

SPECLST_MSLEADER 0.407 0.000 0.492 0.000 1.000 0.381 1.145

SPECLST_MS30 0.540 1.000 0.499 0.000 1.000 -0.161 1.026

BRDSIZE 9.202 9.000 2.299 5.000 19.000 1.101 4.377

BRDNED 0.457 0.455 0.116 0.052 0.778 -0.337 3.466

BRDEXP 0.352 0.333 0.142 0.083 0.857 0.529 2.959

BRDMEET 8.637 8.000 2.584 3.000 21.000 0.972 5.392

ACSIZE 3.635 3.000 0.838 3.000 8.000 1.528 6.015

Variables Mean Median Std. Deviation Minimum Maximum Skewness Kurtosis

ACIND 0.722 1.000 0.448 0.000 1.000 -0.994 1.988

ACEXP 0.388 0.333 0.209 0.000 2.000 1.089 8.109

ACMEET 4.009 4.000 1.212 2.000 10.000 1.483 6.563

INOWN 4.201 0.231 12.221 0.002 147.525 5.076 39.416

BLOCK 23.159 20.64 16.081 0.000 76.120 0.832 3.550

LEVERG 0.640 0.642 0.217 0.056 2.055 0.773 7.215

RETURN 12.852 7.477 16.053 -0.939 179.763 3.448 24.270

SQSUBS 4.702 4.472 2.076 0.000 17.321 1.189 6.434

FORGN 0.524 0.632 0.338 0.000 0.987 -0.358 1.618

LNASSET 6.174 6.129 0.567 4.516 8.122 0.533 3.334

FORGNSALE 45.631 50.205 35.853 0.000 215.62 0.137 2.204

RECINV 0.293 0.264 0.204 0.003 1.022 1.105 4.295

ACQ 1.196 1.000 1.196 0.000 16.000 3.564 22.070

NWFUND 0.922 1.000 0.267 0.000 1.000 -3.169 11.045

NWFUNDRATIO 0.132 0.035 0.500 0.000 9.511 14.757 88.208

NEWDIR 0.472 0.000 0.500 0.000 1.000 0.113 1.013

RESTR 0.409 0.000 0.492 0.000 1.000 0.368 1.135

ROA 9.744 8.34 9.118 -54.44 70.08 0.868 14.755

LNAFEE= the natural log of audit fees; LNNAF=natural log of NAS fees; LNTOTALFEES=natural log of the sum of audit and NAS fees;

FEERATIO1= the fee ratio of NAS fees to total fees; FEERATIO2= the fee ratio of NAS fees to audit fees; SPECLST_MS= continuous

variable which equals to the respective auditor‟ market share; SPECLST_PS= continuous variable which equals to the respective

auditor‟ portfolio share; SPECLST_WEIGHTED= continuous variable which equals to the compliment between auditor‟ market share

(SPECLST_MS) and portfolio share (SPECLST_PS); SPECLST_MSLEADER= coded as 1 if the auditor earned the largest market share

in each particular industry, 0 if otherwise; SPECLST_PSTOP3= coded as 1 if the incumbent auditor earned the highest top three portfolio

shares and 0 if otherwise, and SPECLST_MS30= coded as 1 if the auditor‟ market share exceed 30 percent in each particular industry, 0

if otherwise ; BRDSIZE=the numbers of board‟ members during the year; BDRNED=the proportion of non-executive directors on board

to board size; BRDEXP=the proportion of directors with accounting experience and financial qualification to board size; BRDMEET= the

number of board meetings during the year; ACSIZE= the number of audit committee members; ACIND= coded as 1 if audit committee

had solely non-executive directors; 0 otherwise; ACEXP= the proportion of audit committee members with accounting experience and

financial qualification to audit committee size; ACMEET= the number of audit committee meetings during the year; INOWN=the

cumulative percentage of total shares owned by the directors of a firm; BLOCK= the cumulative percentage shares ownership of the

blockholders who hold at least 5 percent or more of outstanding common shares and who are unaffiiliated with management;

LEVERG=the proportion of debts to total assets; RETURN= the fiscal year total stock return; SQSUBS= the square root of total

consolidated subsidiaries; FORGN=the proportion of foreign subsidiaries to total consolidated subsidiaries; LNASSET= the natural

logarithm of total assets; FORGNSALE= proportion of the firm‟ foreign sales; RECINV= the proportion of total assets in account

receivable and inventory; ACQ= the number of acquisitions made by the company during the year; NWFUND = coded as 1 if the firm

issued new shares or debt for cash during the year, 0 otherwise; NWFUNDRATIO= proceed from new debt and equity issuances/ total

assets; NEWDIR=coded as 1 if the firm appoint new external director during the year, 0 otherwise; RESTR= coded as 1 if the

restructuring program has occurred during the year; ROA= return on assets.

Table 5.2: Pairwise correlation matrix (N=674)

Variable A B C D E F G H I J K

A LNAFEE 1.000

B LNNAF 0.552 1.000

C LNTOTALFEES 0.919 0.756 1.000

D FEERATIO1 -0.113 0.637 0.273 1.000

E FEERATIO2 -0.165 0.413 0.211 0.805 1.000

F SPECLST_MS 0.244 0.140 0.234 0.000 -0.023 1.000

G SPECLST_PS 0.172 0.057 0.128 -0.099 -0.091 0.161 1.000

H SPECLST_WEIGHTED 0.212 0.109 0.183 -0.057 -0.065 0.504 0.883 1.000

I SPECLST_MSLEADER 0.206 0.098 0.190 -0.018 -0.038 0.799 0.395 0.657 1.000

J SPECLST_MS30 0.271 0.147 0.241 -0.039 -0.076 0.776 0.400 0.628 0.764 1.000

K BRDSIZE 0.475 0.310 0.466 0.017 -0.008 0.261 -0.063 0.077 0.231 0.224 1.000

L BRDNED 0.467 0.310 0.453 0.013 -0.021 0.019 0.026 0.015 0.037 0.058 0.062

M BRDEXP -0.167 -0.106 -0.167 -0.011 -0.005 -0.100 0.077 -0.025 -0.144 -0.078 -0.362

N BRDMEET -0.022 -0.019 -0.003 0.033 0.052 -0.110 0.098 0.034 -0.043 -0.027 -0.138

O ACSIZE 0.414 0.207 0.375 -0.054 -0.085 0.145 -0.098 -0.025 0.126 0.131 0.451

P ACIND 0.106 0.084 0.096 -0.012 -0.027 0.039 0.115 0.087 0.027 0.080 -0.012

Q ACEXP -0.089 -0.050 -0.078 0.014 0.039 -0.031 0.089 0.044 -0.052 0.036 -0.066

R ACMEET 0.337 0.271 0.361 0.090 0.069 -0.027 -0.016 -0.022 0.004 -0.040 0.230

S INOWN -0.164 -0.034 -0.120 0.074 0.140 -0.083 -0.007 0.029 -0.043 -0.104 -0.025

T BLOCK -0.172 -0.060 -0.126 0.078 0.134 -0.023 -0.103 -0.092 -0.031 -0.053 0.010

U LEVERG 0.200 0.090 0.178 -0.044 -0.039 0.069 0.105 0.118 0.052 0.099 0.092

V RETURN -0.011 0.015 -0.017 0.006 -0.034 0.444 -0.062 -0.097 -0.082 -0.090 -0.014

W SQSUBS 0.406 0.226 0.364 -0.061 -0.097 0.083 0.227 0.199 0.085 0.122 0.203

X FORGN 0.481 0.253 0.421 -0.113 -0.103 0.084 0.167 0.116 0.072 0.084 0.179

Y LNASSET 0.746 0.458 0.725 0.016 -0.046 0.235 -0.051 0.067 0.192 0.218 0.504

Z FORGNSALE 0.470 0.257 0.425 -0.070 -0.077 0.070 0.189 0.136 0.059 0.120 0.174

AA RECINV -0.172 -0.172 -0.200 -0.099 -0.043 -0.106 0.028 0.002 -0.128 -0.124 -0.155

AB ACQ 0.246 0.111 0.218 -0.055 -0.058 0.062 0.155 0.131 0.082 0.120 0.120

AC NWFUND 0.126 0.053 0.084 -0.078 -0.119 -0.053 -0.016 -0.055 -0.089 -0.066 0.028

AD NWFUNDRATIO -0.030 0.000 -0.008 0.038 0.057 -0.054 -0.087 -0.083 -0.046 -0.072 -0.043

AE NEWDIR 0.065 0.078 0.086 0.060 0.047 0.077 -0.067 -0.024 0.029 0.049 0.098

AF RESTR 0.251 0.216 0.265 0.068 0.014 -0.037 0.003 -0.028 -0.093 -0.006 0.111

AG ROA -0.096 -0.044 -0.103 -0.013 -0.007 0.014 0.025 0.009 0.019 -0.013 -0.056

Variable L M N O P Q R S T U V

L BRDNED 1.000

M BRDEXP 0.003 1.000

N BRDMEET 0.139 0.010 1.000

O ACSIZE 0.363 -0.211 -0.012 1.000

P ACIND 0.446 0.086 0.121 0.023 1.000

Q ACEXP -0.045 0.586 -0.041 -0.229 0.080 1.000

R ACMEET 0.263 -0.036 0.220 0.160 0.024 -0.059 1.000

S INOWN -0.170 -0.025 -0.192 -0.100 -0.091 0.003 -0.133 1.000

T BLOCK -0.121 0.012 -0.093 -0.129 -0.065 0.073 -0.092 0.064 1.000

U LEVERG 0.074 0.024 0.182 0.049 0.087 0.024 0.089 -0.168 -0.072 1.000

V RETURN 0.079 0.089 -0.019 0.036 0.054 0.101 -0.042 -0.005 0.040 0.078 1.000

W SQSUBS 0.070 0.055 -0.028 0.134 0.036 0.004 0.094 -0.055 -0.043 0.089 0.009

X FORGN 0.203 0.019 -0.145 0.127 0.066 0.023 0.148 0.038 0.018 -0.109 0.047

Y LNASSET 0.432 -0.218 0.034 0.416 0.053 -0.115 0.336 -0.149 -0.227 0.059 -0.092

Z FORGNSALE 0.168 -0.023 -0.173 0.139 -0.037 0.056 0.128 0.022 -0.023 -0.140 -0.016

AA RECINV -0.087 0.044 -0.037 -0.098 0.066 -0.017 -0.126 0.014 -0.040 -0.001 0.050

AB ACQ 0.103 -0.011 0.016 0.003 0.099 0.022 -0.046 -0.060 -0.103 0.082 0.022

AC NWFUND -0.008 -0.005 0.035 0.000 -0.043 -0.085 0.030 -0.153 -0.018 0.038 -0.045

AD NWFUNDRATIO -0.019 0.054 0.028 -0.047 0.006 0.043 0.023 -0.000 0.022 -0.035 -0.022

AE NEWDIR 0.147 0.052 -0.015 0.142 0.015 0.057 0.050 -0.057 -0.035 0.031 0.008

AF RESTR 0.131 0.002 -0.024 0.125 0.044 -0.008 0.151 -0.105 -0.039 0.060 0.097

AG ROA -0.075 0.048 -0.060 -0.084 -0.065 0.051 0.022 0.063 0.049 -0.087 -0.024

Variable W X Y Z AA AB AC AD AE AF AG

W SQSUBS 1.000

X FORGN 0.346 1.000

Y LNASSET 0.141 0.119 1.000

Z FORGNSALE 0.298 0.713 0.137 1.000

AA RECINV 0.012 -0.037 -0.260 -0.060 1.000

AB ACQ 0.071 0.112 0.162 0.078 0.028 1.000

AC NWFUND 0.064 0.093 0.103 0.058 -0.004 0.083 1.000

AD NWFUNDRATIO -0.013 0.044 -0.063 0.080 -0.069 -0.003 0.085 1.000

AE NEWDIR 0.032 -0.010 0.077 -0.001 -0.026 0.043 0.039 0.034 1.000

AF RESTR 0.194 0.152 0.150 0.135 -0.011 -0.005 0.083 0.022 0.059 1.000

AG ROA 0.028 0.112 -0.211 0.095 -0.12 -0.050 -0.061 0.102 -0.004 -0.101 1.000 LNAFEE= the natural log of audit fees; LNNAF=natural log of NAS fees; LNTOTALFEES=natural log of the sum of audit and NAS fees; FEERATIO1= the fee

ratio of NAS fees to total fees; FEERATIO2= the fee ratio of NAS fees to audit fees; SPECLST_MS= continuous variable which equals to the respective

auditor‟ market share; SPECLST_PS= continuous variable which equals to the respective auditor‟ portfolio share; SPECLST_WEIGHTED= continuous

variable which equals to the compliment between auditor‟ market share (SPECLST_MS) and portfolio share (SPECLST_PS); SPECLST_MSLEADER= coded

as 1 if the auditor earned the largest market share in each particular industry, 0 if otherwise; SPECLST_PSTOP3= coded as 1 if the incumbent auditor earned

the highest top three portfolio shares and 0 if otherwise, and SPECLST_MS30= coded as 1 if the auditor‟ market share exceed 30 percent in each particular

industry, 0 if otherwise ; BRDSIZE=the numbers of board‟ members during the year; BDRNED=the proportion of non-executive directors on board to board

size; BRDEXP=the proportion of directors with accounting experience and financial qualification to board size; BRDMEET= the number of board meetings

during the year; ACSIZE= the number of audit committee members; ACIND= coded as 1 if audit committee had solely non-executive directors; 0 otherwise;

ACEXP= the proportion of audit committee members with accounting experience and financial qualification to audit committee size; ACMEET= the number of

audit committee meetings during the year; INOWN=the cumulative percentage of total shares owned by the directors of a firm; BLOCK= the cumulative

percentage shares ownership of the blockholders who hold at least 5 percent or more of outstanding common shares and who are unaffiiliated with

management; LEVERG=the proportion of debts to total assets; RETURN= the fiscal year total stock return; SQSUBS= the square root of total consolidated

subsidiaries; FORGN=the proportion of foreign subsidiaries to total consolidated subsidiaries; LNASSET= the natural logarithm of total assets; FORGNSALE=

proportion of the firm‟ foreign sales; RECINV= the proportion of total assets in account receivable and inventory; ACQ= the number of acquisitions made by

the company during the year; NWFUND = coded as 1 if the firm issued new shares or debt for cash during the year, 0 otherwise; NWFUNDRATIO= proceed

from new debt and equity issuances/ total assets; NEWDIR=coded as 1 if the firm appoint new external director during the year, 0 otherwise; RESTR= coded as

1 if the restructuring program has occurred during the year; ROA= return on assets. Correlation in bold are significant at p<0.01, in italic are significant at

p<0.05 and underline at p<0.10.

5.4 ANALYSIS I: AUDIT FEES

Table 5.3 presents the results for the audit fees model for each year and for the pooled

sample. The F-statistics for all models are significant at p<0.01, suggesting that the

models are statistically valid. The adjusted R2 for all models ranges between 77.2%

and 79.5%, showing that the models have a higher explanatory power and this is

consistent with prior studies on audit fees (Simunic, 1980; Abbott et al., 2003a).

As expected the BRDNED is significant and positively related to audit fees across all

models, suggesting the firms with a higher proportion of independent non-executive

directors on board are likely to have higher audit fees. This result confirms the results

from Carcello et al. (2002) and Abbot et al. (2003a) who argued that independent non-

executive directors on boards demand an additional and extensive audit effort in order

to certify their monitoring function, thus increasing the audit fees and the perceived

audit quality, primarily for the purpose of safeguarding their interests. Compared to

the previous UK studies by O‟Sullivan (2000) and Adelopo (2010), the primary

evidence suggests that in terms of „independent‟ characteristics, it seems to be

insensitive in differentiating the types of non-executive director.52

This result is

consistent with prior UK studies.

The other hypotheses variables such as BRDSIZE, BRDEXP, BRDMEET, ACSIZE,

ACIND, ACEXP and ACMEET are either marginally significant in the opposite

predictions or insignificant with audit fees in year-by-year or pooled samples.

BRDSIZE is insignificant with audit fees in all models. However, in the year 2008 and

in the pooled sample, the results contradict those of Carcello et al. (2002). BRDMEET

and BRDEXP are significantly and negatively related to audit fees, suggesting that

boards of directors that are equipped with financial expertise and that have a higher

frequency of board meetings are associated with lower audit fees. These results may

be influenced by the supply-based perspective (Tsui et al., 2001; Krishnan and

Visvanathan, 2009). There is a possibility that auditors value the financial expertise

In O‟Sullivan (2000), he defines the proportion of non-executive directors as the percentage of

outsider members, while in Adelopo (2010) the independent non-executive directors are defined as

those directors with no business or contractual relationships with the firms. Both studies suggest a

positive relationship between non-executive directors and audit fees. In this thesis the „independent‟

characteristic is assigned on the basis of the criteria outlined in the UK Corporate Governance Code

(2010).

and the active involvement of board members, and perceive these characteristics to be

associated with a more effective monitoring function of the board, consequently

reduce their risk assessment, audit effort and fees accordingly.

ACSIZE is significant and positively related to audit fees in the year 2005 at

p<0.10(t=1.78). This weak relationship is also shown in other models and suggests

that there is no evidence that it is linked with audit fees. The other audit committee

characteristics (ACIND, ACMEET, and ACEXP) are insignificant with audit fees in all

models. The insignificant findings on these variables contradict the findings of Abbott

et al. (2003a) but are relatively similar to Carcello et al. (2002). The mixed findings

may be due variation in the nature of the sample selections. Abbott et al. (2003)

examine an overall sample of firms that filed their statements with SEC, which

contains both smaller and larger firms. On the other hand, Carcello et al. (2002)

examine a sample of Fortune 1000 firms which basically contains larger firms than

are found in the sample population examined by Abbott et al. (2003a). The nature of

the sample examined by Carcello et al (2002) was similar to that which has been

analysed in this study. Thus, it is reasonable to expect similar findings to Carcello et

al. (2002).

The results for all the control variables are significant in the predicted directions

except for RECINV, which is insignificant with audit fees across all models. There is a

significant positive relationship between audit fees and the complexity of a firms‟

operations (SQSUBS and FORGN) at p<0.01. As predicted, the firms with a higher

number of consolidated subsidiaries and foreign subsidiaries are likely to have higher

audit fees since the auditors need to put more audit effort and audit hours into dealing

with complex operations, thus increasing the audit fees. These results are consistent

with Simunic (1980), Craswell and Francis (1999), Abbott et al. (2003a) and Carcello

et al. (2002).

LEVERG is positive and significantly related to audit fees at p<0.01, suggesting that

auditors perceive that the firms with higher leverage are associated with a higher

litigation risk, which may lead to more frequent audit failures, due to their poor

financial condition. Thus, an auditor may increase their audit effort and fees for these

firms (i.e. higher leverage firms) in order to compensate their litigation risk. This

result is consistent with the findings of prior studies (Simunic and Stein, 1996; Menon

and Williams, 2001, Pratt and Stice, 1994).

The firm size (LNASSET) is positive and significant with audit fees at p<0.01. This

may suggest that as a firm‟s size increases, auditors enlarge the audit scope and

extend the audit hours, which in turn results in higher audit fees. This result is

consistent with the results from Simunic (1980), Abbott et al. (2003a) and Carcello et

al. (2002).

In summary, the result from the multivariate regression is consistent with the

proposition of agency theory, which suggests that independent non-executive

directors on boards are associated with effective monitoring. They complement their

monitoring function by demanding a higher quality audit from an external auditor in

terms of a more extensive audit effort and a higher number of audit hours, resulting in

higher audit fees and a higher perceived audit quality. The other corporate governance

variables seem to provide inconsistent results or suggest insignificant relationships

with audit fees across the samples by year and within the pooled sample. In particular,

the effect of the monitoring role of an independent board outweighs the other effective

characteristics of a board and audit committee. Therefore, there is no consistent

evidence that board size, the financial expertise and meeting frequency of boards, and

all audit committee characteristics (e.g. size, composition of independent members,

financial expertise and meeting frequency) are associated with increased audit fees.

The results of all the control variables are significant in the predicted directions and

consistent with the prior studies, except for RECINV, which suggest that it has no

relationship to audit fees.

Table 5.3: The result of multivariate regression for audit fees model

LNAFEE = α0 + β

1BRDSIZE + β

2BRDNED + β

3BRDEXP + β4BRDMEET + β

5ACSIZE +

β6ACIND + β

7ACEXP + β

8ACMEET + β

9SQSUBS + β

10FORGN + β

11RECINV +

LEVERG + β13

LNASSET + ε

Variables

Coefficient (t-statistics)

(N=167)

(N=181)

(N=181) 2008 (N=145)

Pooled

(N=674)

Intercept -1.152

(-5.10)***

-0.894

(-3.92)***

-0.807

(-3.72)***

-0.629

(-2.60)***

-0.882

(-8.04)***

BRDSIZE 0.001

(0.07)

(0.73)

(1.43)

-0.101

(-1.00)

(0.72)

BRDNED 0.526

(2.62)***

(1.97)**

(2.16)**

(2.10)**

(4.52)***

BRDEXP -0.110

(-0.68)

-0.174

(-1.04)

-0.130

(-0.91)

-0.285

(-1.70)*

-0.163

(-2.08)**

BRDMEET -0.002

(-0.33)

-0.007

(-1.06)

-0.003

(-0.48)

-0.014

(-1.77)*

-0.007

(-1.95)*

ACSIZE 0.046

(1.78)*

(1.44)

0.0180

(0.78)

-0.010

(-0.41)

(1.72)

ACIND -0.037

(-0.87)

-0.052

(-1.13)

(0.77)

-0.019

(-0.40)

-0.014

(-0.65)

ACEXP 0.092

(0.84)

(0.53)

-0.027

(-0.31)

(0.29)

(0.51)

ACMEET 0.004

(0.32)

(1.53)

(0.32)

(0.20)

(1.39)

SQSUBS 0.038

(4.27)***

(4.45)***

(4.13)***

(4.94)***

(9.02)***

FORGN 0.468

(8.75)***

(7.99)***

(8.21)***

(6.72)***

(16.58)***

RECINV 0.104

(0.233)

(0.91)

(0.37)

-0.143

(-1.62)

(0.60)

LEVERG 0.401

(5.00)***

(5.50)***

(4.09)***

(3.87)***

(9.81)***

LNASSET 0.481

(12.45)***

(10.30)***

(11.46)***

(11.70)***

(23.94)***

Adj. R2

0.783 0.772 0.780 0.795 0.789

*** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10.

5.4.3 The additional analyses and robustness tests

This section further investigates the results obtained in the primary analysis. The

purpose of additional analyses is to provide reasonable assurance that the main

findings are robust to the specifications of various models.53

5.4.3.1 Heteroscedasticity and muticollinearity checks

To confirm whether or not heteroscedasticity exists, the present study uses the

Breush-Pagan or Cook-Weisberg test. If the p-value is significant, then the null

hypothesis, that the variance of the residuals is constant, would be rejected, suggesting

the presence of heteroscedasticity. As can be seen from Table 5.4, the p-value is

insignificant at p<0.10. Therefore, the null hypothesis has to be accepted, indicating

no presence of heteroscedasticity.54

The Pearson correlation matrix in the previous section (Table 5.2) shows that several

variables have largest correlations. LNASSETS has higher correlation with LNAFEE,

BRDSIZE and BRDNED, while BRDEXP has higher correlation with ACEXP. These

higher correlations with total assets may be harmful and thus degrade the influences

of audit fees, board size and proportion of independent board members. The higher

correlations between BRDEXP and ACEXP are expected since audit committee

expertise is a part of board of directors‟ expertise. In order to further investigate

whether these larger correlations may indicate the problem of muticollinearity, the

present study calculates the variance inflation factor (VIF) and tolerance value. The

results are presented in Table 5.5. If the variables have VIF values greater than 10 or

tolerance values lower than 0.10, then they are considered to have multicollinearity

problems (Gujarati, 2003: 339). Since all the variables have VIF values that are

approximately 1.09 to 2.10 and tolerance values that are higher than 0.10 this suggests

that no multicollinearity problem exists.

The additional analyses and the robustness tests are been performed on the pooled sample and

analysed using the OLS regression, unless stated otherwise.

54 Besides the Breush-Pagan test, White test is also been performed and the result relatively similar that

suggest no indication of heteroskedasticity.

5.4.3.2 Different regression estimators

This section provides alternative regression estimators such as robust regression

(iteratively reweighted least squares), least square with clustered robust regression and

quantile regressions.55

The robust regression (iteratively reweighted least squares) and

the least square with clustered robust regression provide better estimations when the

sample contains mild outliers and does not adequately fulfil the OLS assumptions

(Hamilton, 1999; Chen et al., 2003; Adkins and Hill, 2007; Gujarati, 2003), while

quantile regression disregards all the OLS assumptions (Gujarati, 2003). The results

are presented in Table 5.6. As can be seen, the results of these estimators are

consistent with the main finding. There is not much difference between the results of

OLS regression and the other alternative estimators. BRDNED is positively and

significantly related to audit fees at p<0.01. BRDEXP and BRDMEET are also found

to be significantly and negatively related to audit fees across all regression estimators,

including the OLS regression. Similarly, ACSIZE is significantly and positively

related to audit fees. All the control variables are significantly correlated in predicted

directions except that there is no significant correlation between RECINV and audit

fees. This is relatively consistent with the results reported in the primary findings.

Overall, these results may suggest that the main findings reported in Table 5.6 are

robust to various estimators and to the violation of OLS assumptions

5.4.3.3Client size effects

The main findings suggest significant results for BRDNED. There is a possibility that

this result is driven by client size. Thus, several tests are performed to examine the

influence of firm size on BRDNED. Following the example of Carcello et al. (2002),

the present study split the pooled sample into two subsets of data at the median of

LNASSET (proxy for firm size). The first data set is comprised of the firms that have a

LNASSET below the median and this group is identified as the “small firms”. The

second data set is comprised of the firms that have a LNASSET above the median and

this group is identified as the “large firms”. The regressions are re-run separately on

Using the Stata program, the command of rreg was used to perform the robust regression. For the

least square with clustered robust regression, the cluster is based on the industry supersector code. The

cluster option helps to deal with the independent assumption that the errors associated with one

observation are not correlated with errors of other observation. In the current situation, for example, the

firms within each industry supersector will tend to be more like those in the same supersector than the

firms from different industry supersectors from which their errors are not independent.

these two subsets of firms. The results are presented in Table 5.7. Consistent with the

primary finding, BRDNED is positive and significant for “small” and “large” firms.

However, the result of BRDEXP does not hold in the sample of “large firms”. ACSIZE

is insignificant with audit fees in either of the samples. The results of control variables

in both subsets of firms are relatively similar to those that are reported in the primary

findings. In addition to these, the ACEXP and ACMEET are found to be significantly

and positively related to audit fees in the “small” firms but not in the “large” firms.

The Chow test is also performed to test the differences in the regression coefficient

between the “small” and “large” subsets. As can be seen from the Table 5.8, the F-

statistic values of board and audit committee variables are insignificant except for

ACEXP, thus the null hypothesis that the regression coefficients of board and audit

committee variables in both subsets are equal cannot be rejected, suggesting that the

regression coefficients of board and audit committee variables in the “small” and

“large” firms are essentially similar and consistent (except for ACEXP).

In summary, the BRNED appeared to be reasonably consistent across both subsets,

suggesting the significant result on BRNED is not influenced by the client size.

5.4.3.4 New definitions for board and audit committee variables

The primary results suggest that most of the audit committee variables are

insignificant with audit fees, except for ACSIZE. Following Abbott et al. (2003a), the

present study provides new definitions for audit committee variables to see whether

alternative definitions affect the main results. The new definitions are as follows:

(1) ACIND is now in the continuous version, defined as the proportion of

independent non-executive director on audit committee (ACIND1).

(2) ACEXP is defined as a dichotomous variable, ACEXP1, and coded as 1 if audit

committee had at least one director equipped with financial expertise, and 0 if

otherwise.

(3) ACMEET is also defined as a dichotomous variable, ACMEET1, coded as 1 if

an audit committee meeting frequency is more than the sample median, and 0

if otherwise.

Besides the new definitions of audit committee variables, the present study also

provides alternative specifications for board of director variables. Instead of

continuous versions, BRDNED is now defined as a dummy variable, coded as 1 if

60% of the firm‟s directors are independent, and 0 if otherwise. This variable is

known as BRDNED1. Similarly, BRDSIZE is also now defined as a dummy variable;

BRDSIZE1 is coded as 1 if the firm‟s board size is less than sample median, and 0 if

otherwise. These specifications are cited from DeFond et al. (2005). The other

variable‟s descriptions remain unchanged. The results are presented in Table 5.9.

None of the audit committee variables were significant except ACSIZE, which has a

positive relationship with audit fees in the year 2005, and this is consistent with the

primary findings. The result of BRDNED remains significant in every year and in the

pooled regressions, while BRDEXP and BRDMEET are significantly and negatively

related to audit fees in the pooled model only. The control variables remained

quantitatively significant across all models. In summary, the primary findings are

robust to the alternative definitions of board and audit committee variables.

5.4.3.5Additional control variables

In addition to the control variables included in the main model, there are several

variables that are argued to influence the determinants of audit fees. These variables

are return on assets (ROA), liquidity ratio (LIQ), and growth (GROWTH). The present

study tested whether the inclusion of these variables would affect the primary results.

All of these variables are sourced from Datastream. Following Whisenant et al.

(2003) and Lee and Mande (2005), LIQ is defined as the ratio of current assets

divided by current liabilities, while GROWTH represents the growth rate in sales over

the previous fiscal year. In line with Whisenant et al. (2003), ROA and LIQ are

proxies for risk sharing factors, and thus positive relationships are expected between

these variables and audit fees. GROWTH is a proxy for the client size and the larger

the firms are expected to have higher the audit fees due to the increased scope of

audits and audit testing (Whisenant et al., 2003; Lee and Mande, 2005). Hence,

GROWTH is expected to be positively related to audit fees. The results are presented

in Table 5.10. BRDNED is significant in all year-by-year and pooled regressions.

BRDMEET and BRDEXP are found to be negatively and significantly related to audit

fees in the year 2008 and in the pooled models. ACINSIZE is positive and only

significant in the year 2005 model. The results for all control variables are significant

in the predicted directions in all models except for RECINV, ROA and LIQ, which are

insignificantly related to audit fees across all models. GROWTH is positive and

significantly related to audit fees in the year 2006 and in the pooled models. In

general, the main findings reported in Table 5.6 hold, even with the inclusion of these

additional control variables.

5.4.3.6 Endogeneity and two-stage least squares (2SLS) regression

Prior literature suggests that there is a significant relationship between auditing

services and NAS when both are jointly provided by the same auditor (Simunic, 1984;

Palmrose, 1986b). There are two sets of arguments that NAS fees may affect audit

fees or vice versa. The first argument relates to knowledge spillovers that are thought

to reduce the fixed or marginal costs of audits or NAS. Such decreases in the marginal

cost of audits or NAS may then affect the level of audit fees or NAS fees, depending

on the price elasticity of the demand function for audits or NAS (Siminuc, 1984). This

suggests that there will be a positive relationship between audit and NAS fees. The

second argument is that there is a possibility that auditing services may be used as a

“loss-leader” in order to gain a higher profit margin on NAS fees (Hillson and

Kennelley, 1988: 33). In other words, the auditor discounts auditing services in order

to hold on to the lucrative fees of NAS, which in turn suggests that there will be a

negative relationship between audit and NAS fees. Evidences from prior literature

also suggest that board of director and audit committee characteristics may influence

an auditors‟ risk assessment and audit planning, which in turn affects the audit pricing

(Tsui et al., 2001; Boo and Sharma, 2008; Krishnan and Visvanathan, 2009). To

address these issues, the present study first identifies whether the NAS fees or board

of director and audit committee characteristics may suffer from the endogeneity

problem by performing the Durbin-Wu-Hausman test on each of these variables.

Following Larcker and Rusticus (2010) the instrumental variables (IV) are the lagged

values of the endogenous variables.56

The Durbin-Wu-Hausman tests the null

hypothesis that the residual values of LNNAF, BRDSIZE, BDRNED, BRDMEET,

The IV must fulfill the following conditions: (1) outside the regression model, (2) uncorrelated with

regression errors and (3) strongly correlated with endogenous variables. To ensure the IV is valid the

present study estimated the reduced form equations on the first stage of 2SLS regression and examined

the significance level of the endogenous variables. The t-statistic should be at least 3.3 (Adkins and

Hill, 2007: 249-250). All the IVs meet the suggested criterions.

BRDEXP, ACSIZE, ACIND, ACMEET and ACEXP are jointly equal to zero.57

If the

F-statistic is significant, then the null hypothesis would be rejected, suggesting that

endogeneity is present. Table 5.11 present the results of the Durbin-Wu-Hausman test.

Most of the variables suggest insignificant F-statistics except for LNNAF, BDRNED,

BRDEXP and ACSIZE, which confirm the presence of endogeneity since the F-

statistic is significant.

To mitigate the bias caused by endogeneity, the 2SLS regression is performed on

LNNAF, BDRNED, BRDEXP and ACSIZE. The results are presented in Table 5.12.

Compared with the main finding, the results of 2SLS regressions are relatively

consistent, except for the ACSIZE, which is found to be positively related to LNAFE

in most of the 2SLS models. LNNAF is significant and positively related to LNAFEE,

suggesting that the firms with higher audit fees are likely to have higher NAS fees.

The other variables remained unchanged. In summary, the main finding on BRDNED

is that it continues to have positive relationship with LNAFEE, suggesting that the

inference made regarding BRDNED in the main finding is robust to the presence of

endogeneity.

Since the ACIND is a dummy variable, it is changed to a continuous version, ACIND1, which is

defined as the proportion of independent non-executive directors on an audit committee

Table 5.4: Heteroscedasticity test for audit fees model

Breusch-Pagan or Cook-Weisberg Test

H0 = The variance of the residuals is constant

Reject H0 if p-value is significant

chi2(1) = 0.25

Prob > chi2 = 0.6167

Table 5.5: VIF and tolerance value for audit fees model

Variables VIF Tolerance

BRDSIZE 2.10 0.477

BRDNEXP 2.00 0.500

BRDNED 1.91 0.524

LNASSET 1.90 0.527

ACEXP 1.71 0.586

ACSIZE 1.60 0.625

ACIND 1.36 0.732

FORGN 1.28 0.780

ACMEET 1.27 0.786

SQSUBS 1.21 0.825

BRDMEET 1.18 0.844

RECINV 1.09 0.916

LEVERG 1.09 0.917

Mean VIF 1.52

Table 5.6: The results of different estimators for audit fees model (N=674)

Variables

Robust regression Least square regression

with clustered robust Quantile regression

Intercept -0.881

(-7.75)***

-0.883

(-5.45)***

-0.888

(-8.80)***

BRDSIZE 0.003

(0.60)

(0.41)

-0.003

(-0.60)

BRDNED 0.449

(4.40)***

(3.14)***

(3.39)***

BRDEXP -0.201

(-2.42)**

-0.171

(-1.40)*

-0.350

(-4.70)**

BRDMEET -0.006

(-1.73)*

-0.007

(-3.14)**

-0.009

(-2.94)**

ACSIZE 0.017

(1.56)*

(1.77)*

(1.74)*

ACIND -0.015

(-0.70)

-0.014

(-0.68)

-0.006

(-0.33)

ACEXP 0.047

(0.84)

(0.52)

(0.73)

ACMEET 0.007

(0.96)

(1.73)

(0.73)

SQSUBS 0.039

(8.72)***

(8.38)***

(9.33)***

FORGN 0.465

(16.56)***

(9.28)***

(19.79)***

RECINV 0.004

(0.10)

(0.15)

-0.031

(-0.80)

LEVERG 0.385

(9.54)***

(7.12)***

(11.26)***

LNASSET 0.473

(23.25)***

(12.51)***

(27.99)***

Adj. R2/

Pseudo R2 0.780 0.789 0.552

Table 5.7: The results of the multivariate regression of audit fees model for “small” and

“large” firms.

Variables Coefficient (t-statistics)

“Small firms” (N=337) “Large firms” (N=337)

Intercept -0.712

(-3.04)***

-0.596

(-2.43)***

BRDSIZE 0.003

(0.45)

(1.19)

BRDNED 0.402

(3.01)***

(3.59)***

BRDEXP -0.239

(-2.38)**

-0.065

(-0.49)

BRDMEET -0.003

(-0.64)

-0.008

(-1.56)

ACSIZE 0.009

(0.42)

(0.95)

ACIND -0.0203

(-0.70)

-0.008

(-0.26)

ACEXP 0.116

(1.65)

-0.067

(-0.81)

ACMEET 0.018

(1.76)

(0.49)

SQSUBS 0.049

(8.49)***

(4.68)***

FORGN 0.368

(10.14)***

(12.21)***

RECINV 0.014

(-0.27)

(0.73)

LEVERG 0.353

(7.64)***

(5.41)***

LNASSET 0.441

(10.51)***

(10.00)***

Adj. R2

0.647 0.708

Table 5.8: Chow Test

H0 = The regression coefficient on the small and large subsets are essentially the same

F (1, 646) Prob > F

BRDSIZE 0.55 0.460

BRDNED 0.49 0.485

BRDEXP 1.25 0.264

BRDMEET 0.68 0.409

ACSIZE 0.05 0.817

ACIND 0.09 0.768

ACEXP 3.75 0.053

ACMEET 0.75 0.386

Table 5.9: The results of audit fees model for the alternative test variable definitions

Variables

(N=167)

(N=181)

(N=145)

Pooled

(N=674)

Intercept -1.021

(-4.15)***

-0.919

(-4.01)***

-0.660

(-2.50)**

-0.690

(-2.84)***

-0.836

(-6.98)***

BRDSIZE1 0.022

(0.56)

-0.002

(-0.05)

-0.036

(-0.94)

(0.55)

(0.00)

BRDNED1 0.083

(2.05)**

(1.75)*

(2.35)**

(2.00)**

(4.46)***

BRDEXP -0.010

(-0.08)

-0.172

(-1.41)

-0.152

(-1.45)

-0.194

(-1.23)

-0.132

(-2.26)**

BRDMEET -0.002

(-0.28)

-0.007

(-1.20)

-0.002

(-0.27)

-0.012

(-1.60)

-0.005

(-1.74)*

ACSIZE 0.052

(1.72)*

(1.66)

(1.08)

-0.012

(-0.49)

(1.11)

ACIND1 -0.012

(-0.13)

(0.22)

(0.48)

-0.019

(-0.17)

(0.66)

ACEXP1 -0.032

(-0.45)

(0.90)

(0.07)

-0.011

(-0.18)

(0.23)

ACMEET1 0.018

(0.47)

(0.30)

(0.39)

-0.010

(-0.23)

(0.62)

SQSUBS 0.040

(5.05)***

(4.43)***

(3.95)***

(4.96)***

(9.04)***

FORGN 0.473

(8.40)***

(8.69)***

(8.36)***

(6.57)***

(16.68)***

RECINV 0.089

(0.91)

(0.56)

(0.43)

-0.112

(-1.15)

(0.51)

LEVERG 0.373

(4.80)***

(6.56)***

(4.44)***

(4.52)***

(10.70)***

LNASSET 0.494

(12.93)***

(13.56)***

(13.30)***

(15.64)***

(27.99)***

Adj. R2

0.795 0.783 0.793 0.810 0.792

Table 5.10: The results of audit fees model with the additional control variables

Variables

(N=167)

(N=181)

(N=145)

Pooled

(N=674)

Intercept -1.061

(-4.91)***

-0.770

(-3.65)***

-0.855

(-3.88)**

-0.663

(-3.03)**

-0.872

(-8.44)***

BRDSIZE 0.002

(0.19)

(0.68)

(1.33)

-0.012

(-1.11)

(0.60)

BRDNED 0.523

(2.64)***

(2.09)**

(1.91)*

(2.44)**

(4.59)***

BRDEXP -0.107

(-0.64)

-0.239

(-1.39)

-0.146

(-1.22)

-0.242

(-1.67)*

-0.163

(-2.26)**

BRDMEET -0.002

(-0.29)

-0.008

(-1.46)

-0.003

(-0.42)

-0.014

(-1.91)*

-0.007

(-2.09)**

ACSIZE 0.049

(2.17)**

(1.44)

(1.11)

(1.33)

(1.52)

ACIND -0.042

(-1.05)

-0.063

(-1.47)

(0.93)

-0.030

(-0.61)

-0.011

(-0.56)

ACEXP 0.103

(0.91)

(0.74)

-0.015

(-0.22)

(0.11)

(0.70)

ACMEET 0.006

(0.47)

(1.24)

(0.32)

(0.38)

(1.49)

SQSUBS 0.038

(4.98)***

(5.36)***

(4.00)***

(5.32)***

(9.14)***

FORGN 0.467

(8.28)***

(8.26)***

(8.40)***

(5.39)***

(16.01)***

RECINV 0.090

(0.90)

(0.80)

(0.49)

-0.116

(-1.20)

(0.28)

LEVERG 0.381

(4.65)***

(5.97)***

(4.53)***

(4.43)***

(10.49)***

LNASSET 0.467

(11.00)***

(11.63)***

(11.89)***

(13.88)***

(24.79)***

ROA -0.001

(-0.64)

-0.002

(1.02)

(0.53)

(0.80)

LIQ 0.095

(1.65)

-0.997

(-1.43)

(0.01)

(1.38)

(1.13)

GROWTH 0.000

(0.11)

(1.78)*

(1.08)

-0.000

(-0.03)

(1.98)**

Adj. R2

0.804 0.795 0.799 0.817 0.795

** are significant at p<0.05 and *at p<0.10.

Table 5.11: Endogeneity test for audit fees model

Durbin-Wu-Hausman Test

H0 = the residual of LNNAF, BRDSIZE, BDRNED, BRDMEET, BRDEXP, ACSIZE,

ACIND, ACMEET and ACEXP are exogenous

Reject H0 if F-statistic significant

Variable Chi2 (1)

LNNAF 27.432 (p=0.000)

BRDSIZE 1.476 (p=0.224)

BRDNED 7.469 (p=0.006)

BRDEXP 4.157 (p=0.041)

BRDMEET 2.398 (p=0.121)

ACSIZE 6.521 (p=0.010)

ACIND 0.101 (p=0.749)

ACEXP 0.688 (p=0.406)

ACMEET 0.367 (p=0.544)

Table 5.12: The results of 2SLS regression for audit fees model (N=674)

Variables

LNNAF BRDNED BRDEXP ACSIZE

Intercept -0.750

(-6.42)***

-0.788

(-7.66)***

-0.792

(-7.63)***

-0.865

(-8.57)***

LNNAF 0.242

(6.42)***

(3.98)***

(4.27)***

(4.43)

BRDSIZE -0.003

(-0.55)

(1.37)

-0.002

(-0.38)

-0.003

(-0.66)

BRDNED 0.235

(2.20)**

(4.77)***

(4.09)***

(3.03)***

BRDEXP -0.131

(-1.64)

-0.152

(-2.12)**

-0.292

(-2.94)***

-0.170

(-2.37)**

BRDMEET -0.003

(-0.85)

-0.006

(-1.84)*

-0.006

(-1.80)*

-0.006

(-1.74)*

ACSIZE 0.033

(2.73)***

(0.99)

(2.37)**

(3.37)***

ACIND -0.014

(-0.57)

-0.042

(-1.79)*

-0.012

(-0.59)

-0.006

(-0.30)

ACEXP 0.031

(0.59)

(0.49)

(1.47)

(1.14)

ACMEET -0.003

(-0.40)

(0.46)

(1.07)

(1.06)

SQSUBS 0.029

(6.41)***

(8.91)***

(9.04)***

(8.74)***

FORGN 0.381

(12.27)***

(15.04)***

(15.84)***

(15.83)***

RECINV 0.078

(1.52)

(0.93)

(0.91)

(0.90)

LEVERG 0.329

(9.01)***

(10.14)***

(10.71)***

(10.59)***

LNASSET 0.377

(13.88)***

(20.21)***

(21.68)***

(21.51)***

Adj. R2

0.754 0.801 0.803 0.802

** are significant at p<0.05 and *at p<0.10.

5.5 ANALYSIS II: NAS FEES

Table 5.13 presents the results of the NAS fees model by year and in the pooled

samples. All models are estimated using the OLS regression, except for FEERATIO2,

which is regressed using the least square regression with robust standard error,

because the evidence suggests that the FEERATIO2 model is heteroscedastic. The

result of the heteroscedastic test is provided in the additional analyses and robustness

tests section.

FEERATIO1 and FEERATIO2 are regressed only on the pooled model since year-by-

year models lack of significant F-statistics. The adjusted R2 for LNNAF and

LNTOTALFEES models‟ are between 23.5% and 63.8% relatively. The adjusted R2

for LNTOTALFEES is relatively similar to that reported in Ashbaugh et al. (2003).

Compared to LNTOTALFEES, the adjusted R2 for the LNNAF model is relatively

lower and this is partially due to the 18 firms that report zero NAS fees.58

adjusted R2 for FEERATIO1 is 2.2% and FEERATIO2 is 7.5%, which is relatively

lower than that documented in Abbott et al.‟s (2003b) study, which reports that the

adjusted R2 for FEERATIO2 is in between 9.3% and 17.4%.

In general, the regression results of the LNNAF model are consistent with the

LNTOTALFEES model, while the results of FEERATIO1 are relatively similar to

FEERATIO2. BRDSIZE has positive correlation coefficients with all auditor

independence measurements. It is significantly related to LNNAF (in the year 2005

and in the pooled models) and to LNTOTALFEES (in the years 2005 and 2006 and in

the pooled model) but it is insignificant with FEERATIO1 and FEERATIO2. The

positive relationship may suggest that the firms with smaller board size are more

likely to limit NAS purchased as they believe that the higher level of NAS may

compromise auditor independence. The significant results are relatively consistent

with the previous audit quality study documented in Abbott et al (2004), which

The results of multivariate regressions for LNNAF, FEERATIO1 and FEERATIO2 are relatively

similar when we include or exclude the sample firms that report zero NAS fees.

59 Abbott et al. (2003b: 229) show the regression results based on a full sample and two subsamples.

concluded that a smaller board is more effective in controlling the incidence of

restatement.

Contrary to the expectation, BRDNED is found to be positively and significantly

related to LNNAF and LNTOTALFEES in most years and in the pooled models.

However it is insignificant with FEERATIO1 and FEERATIO2. Ashbaugh et al.

(2003) and Larcker and Richardson (2004) suggest that LNNAF and LNTOTALFEES

are better measurements than NAS ratios for capturing the economic importance of

the client to the auditor. The positive relationship of BRDNED and LNNAF to

LNTOTALFEES may imply that independent boards view the joint provision of audit

and NAS as not necessarily compromising audit independence but possibly

broadening the auditors‟ knowledge and improving their judgments, resulting in a

higher audit quality (see Simunic, 1984; Beck et al., 1988a; Arruñada, 1999a; 1999b;

2000; Wallman, 1996; Goldman and Barlev, 1974).

BRDEXPR is insignificant with all the auditor independence measures. Previously,

Lee (2008) documents BRDEXP as composite index is positively related to the

changes in FEERATIO2. Specifically, the composite index is measured as a

dichotomous variable, coded as 1 if more than half NAS committee board members

are independent and at least 27.27% (sample median) of them are financial expertise,

and 0 if otherwise. Lee‟s study, however, does not report the result for BRDNED and

BRDEXP as single variable. BRDMEET is negative and significant only with

LNTOTALFEES (in the year 2008 and in the pooled models), but no statistical

evidence is found to associate it with other measures.

ACSIZE is negatively related to LNNAF (in the year 2008), FEERATIO1 and

FEERATIO2 (both in the pooled model), but there is no evidence that it is associated

with LNTOTALFEES. ACIND is significantly and negatively related to

LNTOTALFEES in the year 2006 but is insignificant with other measures. Previous

studies report ACIND to be negatively and significantly related to FEERATIO2 as

separate variables (Abbott et al., 2003b) and as composite index variables (Abbott et

al., 2003b; Lee and Mande, 2005 and Lee, 2008).60

Likewise, the result indicates no

significant association between ACEXP and all other auditor independence measures;

this is consistent with the findings of Abbott et al (2003b). However Lee (2008)

reports ACEXPR to be significantly related to LNNAF when it is modelled as a

composite index variable.

The present study finds that ACMEET is positive and significant across all auditor

independence measures in the year 2008 and in the pooled models except for the

FEERATIO2 measure. These results suggest that the firms with an active audit

committee are likely to have higher NAS fees, higher total fees and a higher ratio of

NAS fees to total fees. It may be perceived that, where higher levels of NAS are

purchased, an auditor‟s knowledge is broadened thus improving the overall audit

quality. As prior studies have acknowledged, possible benefits flow from the joint

provision of audit and NAS (Simunic, 1984; Beck et al., 1988a; Arruñada, 1999a;

1999b; 2000; Wallman, 1996; Goldman and Barlev, 1974). Previous studies report

ACMEET to be negatively and significantly related to FEERATIO2 and LNNAF when

it is defined as a composite index (Abbott et al., 2003b; Lee and Mande, 2005), but

provide no such evidence when it is modelled as a separate variable (Abbott et al.,

2003b).

In general, the coefficients of the control variables are relatively significant and stable

across the all measures of auditor independence. The coefficient of INOWN is positive

but only significant in the case of LNNAF and FEERATIO1 (in the pooled models),

whereas the coefficient of BLOCK is also positive but significant only in relation to

FEERATIO1 and FEERATIO2. These results are consistent with the findings of Firth

(1997) but contradictory to Abbott et al (2003b) who find negative coefficients for

these two variables. LEVERG is significantly and positively related to

LNTOTALFEES in the years 2005 and 2006, and in the pooled models, but no

evidence of significant relationships are found in the other models. The present study

finds that there is no significant relationship between RETURN and all other measures

Abbott et al. (2003b) and Lee and Mande (2005) define audit committee effectiveness as a dummy

variable, coded as 1 if the audit committee is solely independent and meets at least four times a year.

Lee (2008) defines audit committee effectiveness as a dummy variable as well by taking into

consideration the audit committee‟s expertise variable, coded as 1 if audit committee is solely

independent and at least one-third of them are financial experts.

of auditor independence and this is consistent with the findings of Abbott et al.

(2003b). LNASSET is significantly and positively related to LNNAF and

LNTOTALFEES in the year-by-year and pooled models but there is no evidence to

indicate an association with FEERATIO1 and FEERATIO2.

ACQ is positive and significantly related to LNNAF and LNTOTALFEES but only in

the selected year. NWFUND is conditional to the measures of auditor independence, it

is found to be positively related to LNTOTALFEES (in the year 2007) but negatively

related to FEERATIO1 and FEERATIO2 in the pooled models. NEWDIR is

significantly and positively related to FEERATIO1 and FEERATIO2 but no evidence

associates it with LNNAF and LNTOTALFEES. RESTR is found to be positively and

significantly related to LNNAF, LNTOTALFEES and FEERATIO in most years and in

the pooled regressions. In general, most of the events that are associated with the

demand for the purchase of NAS are in the predicted directions, in agreement with

Firth (1997), except for NWFUND, which has a negative relationship to FEERATIO1

and FEERATIO2 that may suggest that the firms with new shares issued or debt for

cash do not necessarily require professional advice from the auditors

Overall, the multivariate regression finds consistent evidence that the independent

board is positively correlated with the purchase of NAS. Rather than holding the view

that higher levels of NAS impair auditor independence, independent non-executive

directors on boards seem to support the view that a higher NAS provision improves an

auditor‟s judgment and audit quality due to the knowledge spillover effect. The other

hypothesis variables provide inconsistent support for the view that they are linked

with NAS fees. Among the control variables, LNASSET is found to be significantly

and positively related with NAS fees across both the year-by-year and the pooled

samples.

Table 5.13: The results of multivariate regression for NAS fees model

FEE = α0 + β

1BRDSIZE + β

2BRDNED + β

5ACSIZE + β

6ACIND + β

7ACEXP+ β

8ACMEET +

β9INOWN + β

10BLOCK + β

11LEVERG + β

12RETURN +β

13LNASSET + β

14 ACQ + β

15NWFUND + β

16NEWDIR + β

17RESTR

The dependent variable of FEE is measured as follows:

(1) LNNAF; (2) LNTOTALFEES; (3) FEERATIO1; (4) FEERATIO2

Variable

2005 (N=167) 2006 (N=181) 2007 (N=181) 2008 (N=145) Pooled (N=674)

(1) (2) (1) (2) (1) (2) (1) (2) (1) (2) (3) (4)

Intercept -1.682

(-2.27)**

-0.644

(-1.79)*

-0.580

(-0.87)

-0.116

(-0.38)

-0.780

(-1.04)

-0.432

(-1.33)

-0.863

(-1.16)

-0.211

(-0.60)

-0.776

(-2.24)**

-0.336

(-2.10)**

(3.18)***

(1.33)

BRDSIZE 0.069

(2.03)**

(1.67)*

(0.74)

(1.91)*

(0.62)

(1.41)

(1.54)

(1.50)

(2.37)**

(3.37)***

(0.39)

(0.28)

BRDNED 0.973

(1.54)

(2.82)***

(1.96)*

(2.70)***

(2.06)**

(2.81)***

(1.33)

(2.66)***

(3.32)***

(5.54)***

(0.37)

(0.46)

BRDEXP 0.505

(1.00)

(0.56)

-0.441

(-0.91)

-0.185

(-0.84)

-0.193

(-0.43)

-0.132

(-0.69)

(0.85)

(0.60)

(0.00)

-0.026

(-0.24)

-0.023

(-0.30)

-0.315

(-0.71)

BRDMEET 0.001

(0.04)

-0.007

(-0.62)

-0.012

(-0.62)

-0.007

(-0.85)

-0.006

(-0.25)

-0.013

(-1.33)

-0.029

(-1.36)

-0.020

(-1.97)*

-0.015

(-1.45)

-0.011

(-2.34)**

(1.24)

(1.60)

ACSIZE -0.037

(-0.45)

(0.31)

-0.027

(-0.33)

(0.93)

-0.045

(-0.62)

(0.60)

-0.118

(-1.67)*

-0.050

(-1.49)

-0.055

(-1.49)

(0.25)

-0.024

(-2.06)**

-0.143

(-1.93)*

ACIND -0.094

(-0.71)

-0.016

(-0.25)

-0.095

(-0.73)

-0.109

(-0.183)*

(0.01)

-0.030

(-0.54)

(1.57)

(0.64)

-0.003

(-0.05)

-0.031

(-1.05)

-0.007

(-0.36)

-0.081

(-0.55)

ACEXP 0.021

(0.06)

(0.39)

-0.044

(-0.13)

(0.11)

-0.077

(-0.28)

-0.008

(-0.07)

-0.239

(-0.72)

-0.097

(-0.62)

-0.056

(-0.37)

(0.14)

(0.01)

(0.58)

ACMEET 0.032

(0.69)

(1.48)

(1.07)

(1.63)

(0.89)

(1.54)

(2.06)**

(2.52)**

(2.88)***

(3.88)***

(1.84)*

(1.35)

INOWN 0.005

(1.06)

(1.20)

(0.87)

-0.001

(-0.29)

(0.86)

(1.40)

(0.97)

(0.15)

(1.77)*

(1.21)

(1.91)*

(1.63)

BLOCK 0.001

(0.18)

(1.02)

(0.90)

(1.14)

(1.46)

(0.73)

-0.001

(-0.22)

-0.000

(-0.09)

(1.00)

(1.51)

(1.95)*

(2.54)**

Variable

2005 (N=167) 2006 (N=181) 2007 (N=181) 2008 (N=145) Pooled (N=674)

(1) (2) (1) (2) (1) (2) (1) (2) (1) (2) (3) (4)

LEVERG 0.227

(0.90)

(2.74)***

(1.55)

(3.38)***

-0.019

(-0.07)

(1.06)

(0.18)

(1.52)

(1.51)

(4.74)***

-0.039

(-1.07)

-0.126

(-0.57)

RETURN 0.001

(0.32)

-0.000

(-0.22)

(0.83)

(0.15)

(0.23)

(0.15)

-0.001

(-0.23)

(0.56)

(0.66)

-0.000

(-0.09)

(0.20)

-0.003

(-1.27)

LNASSET 0.402

(3.32)***

(7.13)***

(3.06)***

(6.58)***

(3.27)***

(7.84)***

(3.65)***

(7.69)***

(6.60)***

(15.19)***

(0.67)

-0.026

(-0.20)

ACQ 0.012

(0.40)

(1.34)

(2.41)**

(3.73)***

-0.003

(-0.13)

(2.07)**

-0.014

(-0.54)

(0.86)

(1.18)

(4.23)***

-0.004

(-0.87)

-0.014

(-0.75)

NWFUND 0.441

(2.20)**

(0.77)

-0.269

(-1.51)

-0.096

(-1.18)

(0.78)

(2.972)***

-0.226

(-1.15)

-0.124

(-1.33)

(0.31)

(0.33)

-0.058

(-1.99)**

-0.510

(-1.68)*

NEWDIR 0.007

(0.06)

(0.44)

(1.24)

-0.001

(-0.01)

-0.084

(-0.79)

(0.61)

(0.45)

-0.075

(-1.51)

(0.59)

-0.003

(-0.15)

(1.79)*

(1.73)*

RESTR 0.237

(2.10)**

(2.60)**

(4.24)***

(4.87)***

(0.63)

(1.48)

(0.51)

(1.69)*

(3.70)***

(5.25)***

(1.90)*

(1.16)

Adj. R2 0.235 0.559 0.311 0.638 0.147 0.598 0.317 0.653 0.260 0.614 0.022 0.075

*** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10. All models are been estimated using the OLS regression except for

FEERATIO2 which been regressed using the least square regression with robust standard error because the evidence suggest that the FEERATIO2 model

is heterscedastic.

In this section the present study examines whether the primary findings hold in the

case of the violation of OLS assumptions, and whether they are robust in various other

model specifications. The tests include heteroscedasticity and multicollinearity

checks, different regression estimators, client size analysis, alternative definitions of

hypotheses variables, additional control variables, and an endogeneity test and 2SLS

regression.

5.5.3.1 Heteroscedasticity and muticollinearity checks

Table 5.14 presents the results of the heteroscedasticity test according to the specific

dependent variables. Based on the Breusch-Pagan or Cook-Weisberg test, most

models indicate an insignificant p-value, suggesting that the variances of the residuals

are homogeneous, except for FEERATIO2, which has a significant p-value at p<0.01,

which indicates the present of heteroscedasticity.61

The results of VIF test and tolerance value are presented in Table 5.15.62

None of the

variables had a VIF value of more than 10 or a tolerance value lower than 0.10,

suggesting no indication of a multicollinearity problem.

This section provides the results of multivariate regression using different estimators.

Previously, LNNAF, LNTOTALFEES and FEERATIO1 were regressed using the OLS

estimator, while FEERATIO2 was regressed using the least square regression with

robust standard errors. In this section, LNNAF, LNTOTALFEES and FEERATIO1 will

be regressed using robust regression and quantile regression, while FEERATIO2 will

be regressed using the OLS and GLS regressions. The results are presented in Table

The White test also revealed a similar result. There was no indication of heteroscedasticity in any of

the auditor independence measurements except for the FEERATIO2 model.

62 This result is based on the modelling of LNNAF as a dependent variable. The other models indicate

relatively similar VIF and tolerance values.

The robust regression is efficient when the models contain mild outliers and do not

fulfil the normality assumption (Hamilton, 1999; Chen et al., 2003; Adkins and Hill,

2007; Gujarati, 2003). The quantile regression is one of the non-parametric tests

which do not require any assumptions (Gujarati, 2003). The GLS regression is an

alternative to the least square regression with robust standard error that is efficient in

controlling heteroscedastic models.

As can be seen, the results of robust and quantile regressions are relatively consistent

with the OLS regression presented in the main findings (Table 5.6) except for the

relationship between LNNAF and INOWN and the relationship between FEERATIO1

and BLOCK. Both of these relationships are insignificant in the quantile regression.

The hypothesis variables remain unchanged.63

For the FEERATIO2 model, GLS regression provides consistent results with the least

square regression with robust standard error, as shown in the main findings (Table

5.6). This may be due to the efficiency of both estimators in controlling error

variances so that each sample observation would have a constant variance. However,

the results of OLS regression were slightly different. The main findings and those of

the GLS regression (pooled sample) suggest that ACMEET and INOWN are

insignificantly related with FEERATIO2 but, under OLS regression, both are found to

be positively and significantly related to FEERATIO2 at p<0.10 and p<0.01

respectively. The results for the other variables are relatively consistent with the main

findings. The inconsistency of the results, for variables ACMEET and INOWN under

the OLS regression, may be partly due to the presence of the heteroscedasticity

problem identified in the FEERATIO2 model. The statisticians suggests that in the

presence of heteroscedasticity, the least square estimator with robust standard error

(Huber-White standard errors) or GLS regression are much more efficient than the

OLS estimator because they able to reweight the error variances and thus correct for

heteroscedasticity and autocorrelation (Adkins and Hill, 2007: 196; Gujarati, 2003:

387). In other words, in the presence of heteroscedasticity, the results of OLS

regression may be distorted and biased.

They might be different in terms of levels of significance but this does not change the content of the

main findings.

5.5.3.3Client size effects

The present study re-regress the NAS fees model to see whether the results of

BRDNED are driven by client size. Following the approach of Carcello et al. (2002),

the pooled samples are split into two groups at the median of LNASSET. Firms with

LNASSET below median are grouped together as the “small” firms and those with

LNASSET above median are grouped together as the “large” firms. The regressions

are re-run separately using the OLS estimator.64

The results are presented in Table

5.17. Consistent with the primary findings (pooled model), BRDNED is significantly

and positively related to LNNAF and LNTOTALFEES for both “small” and “large”

firms. The other hypotheses variables and the control variables are either significant in

“small” or “large” firms or significant/insignificant in both groups, depending on the

auditor independence measures. In summary, BRDNED is the only hypotheses

variable that is found to be consistent across “large” and “small” firms and thus it

confirms the main finding that the independent boards is not driven by client size.

As with the audit fees model, the present study tests whether alternative specifications

for hypotheses variables affect the main analysis. Following to approaches of Abbott

et al. (2003a) and DeFond et al. (2005), the alternative definitions for board and audit

committee variables (BRDSIZE, BRDNED, ACIND, ACEXP, ACMEET) are as

follows:

(1) BRDSIZE1is coded as 1 if the firm‟s board size is less than sample median,

and 0 if otherwise.

(2) BRDNED1 is coded as 1 coded as 1 if 60% of the firm‟s directors are

independent, and 0 if otherwise.

(3) ACIND1is defines as the proportion of independent non-executive director on

the audit committee.

(2) ACEXP1 is coded as 1 if audit committee had at least one director equipped

with financial expertise, and 0 if otherwise.

(3) ACMEET1 is coded as 1 if audit committee meeting frequency is more than

the sample median, and 0 if otherwise.

The FEERATIO2 model revealed an insignificant F-statistic (regressed using the least square

regression with robust standard error), suggesting that the model is statistically invalid. Thus, the

results are not reported.

The definitions of the other variables remain unchanged. Table 5.18 presents the

results for the new definitions.65

As can be seen, the results for the alternative

definitions are relatively consistent with the main findings, except for BRDSIZE1 and

ACEXP1. BRDSIZE1 is found to be negatively related to NAS fees measurements,

suggesting the firms with a board size that is smaller than the sample median are

likely to have higher NAS fees. This may indicate that the firms with fewer board

members are likely to have more NAS. The results for the NAS fees variables seem to

be sensitive to the alternative definition of board size. BRDNED1 and ACMEET1 are

positively related to NAS fees, suggesting that the firms with boards that have more

than 60% independent membership, and whose audit committee meetings are more

frequent than the sample median, are likely to have higher NAS fees. ACEXP1 is

negatively related to all auditor independence measurements. This suggesting that the

audit committees with at least one member equipped with financial expertise are more

likely to limit the level of NAS purchased as they have the perception that higher

NAS impairs auditor independence. The results from the control variables are

relatively unchanged. Overall, the primary findings on the independence of boards are

not modified by alternative definitions for independent non-executive directors.

5.4.3.5Additional control variables

Several control variables are included in the primary model to see whether the

inclusion of these variables affect the results. As with the audit fees model, these

variables are: return on assets (ROA), liquidity ratio (LIQ), and growth (GROWTH).

These are predicted to be positively related to NAS fees. Prior literature suggests that

more profitable firms and higher growth firms are likely to have more resources to

purchase NAS (Habib and Islam, 2007; Antle et al., 2006). The results are presented

in Table 5.19. The results for the hypotheses variables and the main control variables

are relatively similar to the main findings. ROA is positively related to

LNTOTALFEES at p<0.10, but insignificantly related with other measures. GROWTH

is found to be positively related to FEERATIO1 and FEERATIO2. These results are

relatively consistent with the findings of Habib and Islam (2007) who suggest that

firms with a higher profitability and higher growth are likely to purchase more NAS. 65

All the models are regressed using the OLS except for FEERATIO2 model, which is estimated using

the least square regression with robust standard error.

Overall, the main finding is unchanged and the additional control variables are

unlikely to affect the results

As previously highlighted in the section of additional analysis and robustness tests for

the audit fees model, evidence from prior studies suggests that there are two possible

outcomes of the joint provision of audit and NAS. The first possibility is related to the

knowledge spillover effect that suggests that there will be a positive relationship

between audit fees and NAS fees. . The second possibility is that higher NAS fees are

used to discount auditing services in order to gain a higher profit margin on the

lucrative NAS fees. This leads to a negative relationship between audit and NAS fees.

Both arguments lead to the issue of endogeneity. In addition, prior literature also

suggests that board of director and audit committee characteristics are likely to be

associated with endogeneity (e.g. Larcker and Richardson, 2004; Larcker and

Rusticus, 2010). Therefore, to investigate these concerns, the present study first

examined whether these variables suffer from endogeneity by performing the Durbin-

Wu-Hausman test.66

Table 5.11 presents the results of the Durbin-Wu-Hausman test

for each variable analysed under LNNAF, LNTOTALFEES, FEERATIO1 and

FEERATIO2 models. As can be seen, the F-statistics are insignificant in all models

suggesting no indication of endogeneity. This may not required the 2SLS test. As

suggested by Baum et al. (2003), in the absence of endogeneity, the results of 2SLS

regressions are unacceptable and biased. In summary, the results estimated using the

least square regression with robust standard error in the main analysis is more

efficient due to the absence of endogeneity.

As with the audit fees model, ACIND is changed to a continuous version of the variable and is called

ACIND1 (defined as the proportion of independent non-executive directors on the audit committee) in

order to perform the Durbin-Wu-Hausman test.

Table 5.14: Heteroscedasticity test for NAS fees model

Dependent variable chi2(1) Prob > chi2

LNNAF 0.27 0.606

LNTOTALFEES 0.47 0.492

FEERATIO1 1.62 0.203

FEERATIO2 417.92 0.000

Table 5.15: VIF and tolerance values for NAS fees model

BRDSIZE 2.11 0.474

LNASSET 2.02 0.495

BRDNED 2.01 0.498

BRDEXP 1.89 0.531

ACEXP 1.73 0.579

ACSIZE 1.65 0.605

ACIND 1.37 0.730

ACMEET 1.30 0.771

BRDMEET 1.19 0.843

INOWN 1.12 0.890

BLOCK 1.11 0.898

ACQ 1.09 0.915

LEVERG 1.09 0.916

RESTR 1.07 0.932

NWFUND 1.07 0.937

RETURN 1.06 0.940

NEWDIR 1.06 0.943

Mean VIF 1.41

Table 5.16: The results of different estimators for NAS fees model (N=674)

Variable

Robust regression Quantile regression OLS

regression

LNNAF LNTOTALFEES FEERATIO1 LNNAF LNTOTALFEES FEERATIO1 FEERATIO2 FEERATIO2

Intercept -0.855

(-3.57)***

-0.348

(-2.17)**

(2.82)***

-0.756

(-3.49)***

-0.235

(-1.22)

(2.43)**

(1.55)

(1.26)

BRDSIZE 0.025

(2.33)**

(3.44)***

(0.34)

(2.27)**

(2.46)**

(0.57)

(0.31)

(0.21)

BRDNED 0.693

(3.40)***

(5.35)***

(0.20)

(3.84)***

(4.56)***

(0.36)

(0.48)

(0.50)

BRDEXP -0.086

(-0.53)

-0.033

(-0.31)

-0.015

(-0.19)

-0.042

(-0.29)

-0.132

(-1.02)

-0.017

(-0.22)

-0.315

(-0.66)

-0.306

(-0.69)

BRDMEET -0.006

(-0.92)

-0.010

(-2.08)**

(1.14)

-0.018

(-1.38)

-0.012

(-2.06)**

(0.80)

(1.05)

(1.59)

ACSIZE 0.000

(0.01)

(0.40)

-0.023

(-1.90)*

(0.53)

(1.29)

-0.026

(-1.98)**

-0.143

(-1.89)*

-0.151

(-1.99)**

ACIND -0.022

(-0.50)

-0.033

(-1.11)

-0.005

(-0.22)

-0.019

(-0.48)

-0.055

(-1.58)

-0.018

(-0.80)

-0.081

(-0.63)

-0.097

(-0.64)

ACEXP 0.002

(0.02)

-0.013

(-0.18)

-0.001

(-0.03)

(0.14)

(0.55)

-0.037

(-0.75)

(0.66)

(0.54)

ACMEET 0.06

(3.85)***

(3.75)***

(1.82)*

(2.51)**

(1.67)*

(1.93)*

(1.29)

INOWN 0.003

(1.73)*

(0.22)

(1.65)*

(1.15)

(0.09)

(1.64)*

(3.44)***

(1.59)

BLOCK 0.002

(1.02)

(1.51)

(1.79)*

(0.66)

(1.22)

(1.45)

(1.45)***

(2.43)**

LEVERG 0.164

(1.43)

(4.91)***

-0.043

(-1.10)

(1.41)

(4.10)***

-0.045

(-1.11)

-0.126

(-0.53)

-0.103

(-0.46)

Variable

Robust regression Quantile regression OLS

regression

LNNAF LNTOTALFEES FEERATIO1 LNNAF LNTOTALFEES FEERATIO1 FEERATIO2 FEERATIO2

RETURN -0.000

(-0.27)

-0.000

(-0.08)

(0.30)

(0.20)

(0.14)

(0.44)

-0.003

(-0.99)

-0.003

(-1.28)

LNASSET 0.445

(10.64)***

0.4259

(15.28)***

(0.87)

(12.085)***

(12.20)***

(1.21)

-0.026

(-0.21)

-0.015

(-0.11)

ACQ 0.021

(2.39)**

(4.13)***

-0.003

(-0.73)

(4.03)***

(4.32)***

(0.30)

-0.014

(-0.53)

-0.014

(-0.72)

ISSUE -0.088

(-1.36)

(0.01)

-0.068

(-2.16)**

(1.32)

(0.23)

-0.077

(-2.38)**

-0.510

(-2.69)***

-0.536

(-1.72)*

NEWDIR 0.035

(1.02)

(0.07)

(1.78)*

-0.010

(-0.33)

(0.90)

(1.67)*

(1.72)*

(1.75)*

REST 0.131

(3.72)***

(5.12)***

(1.83)*

(3.58)**

(3.99)***

(2.25)**

(1.07)**

(1.15)

Adj. R2/ Pseudo

R2 0.295 0.618 0.018 0.418 0.409 0.024 0.051 0.052

Table 5.17: The results of NAS fees model for “small” and “large” firms (N=337)

Variables

LNNAF LNTOTALFEES FEERATIO1

“Small” “Large” “Small” “Large” “Small” “Large”

Intercept -0.010

(-0.01)

-1.013

(-1.28)

-0.015

(-0.04)

-0.392

(-1.14)

(2.31)**

(0.96)

BRDSIZE 0.42

(2.03)**

(1.24)

(2.82)***

(2.08)**

(0.06)

(0.92)

BRDNED 1.337

(3.30)***

(1.66)*

(4.14)***

(3.73)***

(1.05)

-0.021

(-0.16)

BRDEXP 0.019

(0.06)

(0.05)

-0.236

(-1.56)

(1.49)

(0.15)

-0.059

(-0.55)

BRDMEET 0.001

(0.07)

-0.029

(-1.82)*

-0.001

(-0.19)

-0.021

(-3.11)***

(1.11)

(0.77)

ACSIZE -0.149

(-2.25)**

-0.029

(-0.59)

-0.052

(-1.67)*

(0.83)

-0.059

(-2.57)**

-0.011

(-0.79)

ACIND -0.149

(-1.70)*

(1.02)

-0.041

(-0.961)***

-0.041

(-0.99)

-0.056

(-1.84)*

(1.14)

ACEXP -0.011

(-0.05)

-0.076

(-0.34)

(1.03)

-0.087

(-0.90)

-0.016

(-0.22)

(0.59)

ACMEET 0.039

(1.25)

(2.53)**

(2.68)***

(2.65)***

(0.10)

(2.47)**

INOWN 0.001

(0.43)

(1.91)*

-0.000

(-0.34)

(1.96)*

(0.47)

(1.81)*

BLOCK 0.002

(0.93)

(0.48)

(1.01)

(1.18)

(0.96)

(1.95)*

LEVERG 0.145

(1.03)

(0.74)

(2.82)***

(3.05)***

-0.042

(-0.85)

-0.049

(-0.76)

RETURN 0.001

(0.58)

(0.63)

(0.18)

(0.32)

(0.22)

(0.34)

LNASSET 0.312

(2.40)**

(3.20)***

(6.29)***

(7.44)***

(0.10)

-0.001

(-0.04)

ACQ 0.025

(1.06)

(0.79)

(2.78)***

(3.27)***

(0.11)

-0.005

(-1.08)

NWFUND -0.064

(-0.57)

(0.96)

-0.028

(-0.51)

(0.91)

-0.119

(-3.09)***

(0.79)

NEWDIR 0.029

(0.43)

(0.37)

-0.023

(-0.73)

(0.64)

(1.39)

(0.84)

REST 0.162

(2.30)**

(2.82)***

(3.39)***

(4.24)***

(1.85)*

(0.50)

Adj. R2

0.080 0.190 0.309 0.504 0.027 0.027

Table 5.18: The results of NAS fees model for the alternative test variable definitions

(N=674)

Variables

LNNAF LNTOTALFEES FEERATIO1 FEERATIO2

Intercept -0.407

(-1.01)

-0.043

(-0.23)

(3.69)***

(1.87)*

BRDSIZE1 -0.080

(-1.70)*

-0.064

(-2.30)**

(1.41)

-0.208

(-1.69)*

BRDNED1 0.121

(2.11)**

(4.31)***

-0.011

(-0.63)

-0.165

(-1.41)

BRDEXP 0.073

(0.37)

(0.65)

(0.83)

(0.64)

BRDMEET -0.012

(-1.21)

-0.009

(-1.92)*

(1.47)

(1.95)*

ACSIZE -0.023

(-0.66)

(1.27)

-0.022

(-2.20)**

-0.139

(-2.22)**

ACIND1 0.016

(0.18)

-0.025

(-0.58)

(0.11)

-0.011

(-0.09)

ACEXP1 -0.184

(-1.77)*

-0.100

(-2.06)**

-0.093

(-2.81)***

-0.386

(-2.11)**

ACMEET1 0.152

(2.44)**

(2.71)***

(1.79)*

(1.39)

INOWN 0.003

(1.43)

(0.82)

(1.35)

(1.49)

BLOCK 0.002

(1.38)

(1.98)*

(2.31)**

(2.77)***

LEVERG 0.197

(1.67)*

(4.66)***

-0.030

(-0.86)

-0.075

(-0.33)

RETURN 0.001

(0.63)

-0.000

(-0.03)

(0.14)

-0.003

(-1.19)

LNASSET 0.493

(8.92)***

(18.84)***

(1.22)

(0.30)

ACQ 0.018

(1.40)

(4.23)***

-0.003

(-0.79)

-0.010

(-0.50)

NWFUND -0.004

(-0.05)

-0.008

(-0.18)

-0.072

(-2.09)**

-0.613

(-2.03)**

NEWDIR 0.038

(0.76)

(0.16)

(1.79)*

(1.77)*

REST 0.198

(3.85)***

(5.43)***

(2.00)**

(1.22)

Adj. R2

0.244 0.598 0.034 0.062

Table 5.19: The results of NAS fees model with the additional control variables (N=674)

Variables Coefficient (t-statistics)

LNNAF LNTOTALFEES FEERATIO1 FEERATIO2

Intercept 0.923

(-2.55)**

-0.434

(-2.61)***

(3.14)***

(1.43)

BRDSIZE 0.036

(2.32)**

(3.26)***

(0.59)

(0.50)

BRDNED 0.958

(3.22)***

(5.50)***

(0.53)

(0.69)

BRDEXP 0.001

(0.00)

-0.028

(-0.26)

-0.020

(-0.27)

-0.313

(-0.72)

BRDMEET -0.014

(-1.37)

-0.010

(-2.21)**

(1.25)

(1.64)

ACSIZE -0.055

(-1.47)

(0.30)

-0.025

(-2.10)**

-0.147

(-1.98)**

ACIND 0.002

(0.04)

-0.027

(-0.94)

-0.009

(-0.44)

-0.096

(-0.65)

ACEXP -0.067

(-0.44)

(0.01)

-0.009

(-0.19)

(0.42)

ACMEET 0.061

(2.67)***

(3.54)***

(1.70)*

(1.27)

INOWN 0.004

(1.69)*

(1.06)

(1.67)*

(1.49)

BLOCK 0.002

(0.99)

(1.56)

(1.84)*

(2.44)**

LEVERG 0.193

(1.64)

(4.95)***

-0.0033

(-0.89)

-0.076

(-0.35)

RETURN 0.001

(0.71)

(0.09)

(0.26)

-0.003

(-1.16)

LNASSET 0.418

(6.74)***

(15.26)***

(0.51)

-0.056

(-0.42)

ACQ 0.015

(1.19)

(4.06)***

-0.004

(-0.96)

-0.018

(-0.95)

NWFUND 0.032

(0.34)

(0.27)

-0.062

(-2.10)**

-0.547

(-1.82)*

NEWDIR 0.028

(0.55)

-0.003

(-0.15)

(1.72)*

(1.65)*

REST 0.194

(3.80)***

(5.40)***

(1.86)*

(1.09)

ROA 0.004

(1.35)

(1.85)*

-0.001

(-0.74)

-0.007

(-1.22)

LIQ -0.046

(-0.48)

(0.42)

-0.027

(-0.88)

-0.154

(-0.80)

GROWTH 0.001

(0.45)

(1.00)

(1.99)**

(2.07)**

Adj. R2

0.260 0.616 0.024 0.058

Table 5.20: Endogeneity test for NAS fees model

H0 = the residual of LNNAF, BRDSIZE, BDRNED, BRDMEET, BRDEXP, ACSIZE,

ACIND1, ACMEET and ACEXP are exogenous

Variable LNNAF LNTOTALFEES FEERATIO1 FEERATIO2

Chi2 (1)

LNNAF 2.544 (p=0.110) 0.361 (p=0.548) 2.544 (p=0.110) 0.249 (p=0.617)

BRDSIZE 0.188 (p=0.664) 0.191 (p=0.661) 0.188 (p=0.664) 0.021 (p=0.884)

BRDNED 1.492 (p=0.221) 0.133 (p=0.715) 1.492 (p=0.221) 0.076 (p=0.782)

BRDEXP 0.250 (p=0.614) 0.088 (p=0.766) 0.254 (p=0.614) 0.001 (p=0.965)

BRDMEET 0.136 (p=0.711) 0.106 (p=0.744) 0.136 (p=0.711) 0.092 (p=0.760)

ACSIZE 0.016 (p=0.898) 0.930 (p=0.335) 0.016 (p=0.898) 0.269 (p=0.603)

ACIND1 1.299 (p=0.254) 0.185 (p=0.667) 1.299 (p=0.254) 0.188 (p=0.664)

ACEXP 0.101 (p=0.750) 0.026 (p=0.870) 0.101 (p=0.750) 0.031 (p=0.859)

ACMEET 0.205 (p=0.650) 1.069 (p=0.301) 0.205 (p=0.650) 1.026 (p=0.310)

5.6 ANALYSIS II1: INDUSTRY SPECIALIST AUDITOR

5.6.1 Univariate test

Table 5.21 presents the results of mean differences tests for all the variables in the

industry specialist model according to the dichotomous definitions of SPEC_AUD.

Based on the SPECLST_MSLEADER and the SPECLST_30MS definitionss, industry

specialists are employed in 274 and 364 firm-years, while non-specialist auditors are

employed in 400 and 310, respectively. In general, the results of the t-test using the

SPECLST_30MS definition show more significant differences between the groups of

auditors in the hypothesis and control variables than using the SPEC_MSLEADER

definition.

The results of the t-tests of both definitions show that the firms with the larger board

size tend to employ industry specialists rather than non-specialist auditors, while the

boards with higher levels of financial expertise are more likely to appoint a non-

specialist auditor. However, there are no significant differences between the two

groups of auditors for firms that have higher percentages of independent board

members and higher board meeting frequencies.

In regard to audit committee characteristics, the results of the t-tests suggest that there

are significant differences in audit committee size between the two groups of auditors.

Both dichotomous definitions suggest that the larger audit committees are more likely

to appoint industry specialists than they are to appoint non-specialist auditors. The

result for independent audit committees is conditional. Under the

SPECLST_MSLEADER definition, the results for independent audit committees show

no significant difference between these two groups of auditors. However, using the

SPECLST_30MS definition, the likelihood of firms with solely independent audit

committees is higher in the group that employ industry specialist auditors than in the

group that employ non-specialist auditors. These findings are consistent with Chen et

al. (2005). However, when either definition is used, audit committee expertise and

audit committee meeting frequencies do not appear to be associated with any

significant difference between the group of firms using industry specialists and the

group of firms using non-specialists.

The means that the levels of LNASSETS, LEVERGN, SQSUBS and FORGNSALE are

higher for firms with specialist auditors, and this is consistent with the argument that:

the higher a firms‟ risk and the higher the complexity of firm operations, the more the

need for a higher quality auditor (Collier and Gregory, 1996; Simon and Francis,

1988). In addition, the result also suggests that the firms with higher percentages of

total shares owned by the directors (INOWN) are more likely to employ non-

specialists, and this is consistent with the report found in Abbott and Parker (2000).

The results for the IND_SPEC model are presented in Table 5.22, by year and by

pooled samples.67

However, SPECLST_MS, SPECLST_PS and

SPECLST_WEIGHTED are modelled only on the pooled sample due to the lack of F-

statistics for several years. The adjusted R2s, or pseudo-R

2s, for all models are

between 6.2% and 15.9%, and these values are relatively higher than those reported in

Abbott et al.‟ (2003b) and Chen et al.‟ (2005), who report them to be in between 2 %

and 10 %, and 6.7% and 7.6 % respectively.

In general, the regression results for SPEC_AUD, which is measured using the market

share approach, are relatively consistent across both the year-by-year and the pooled

samples. In contradiction to the expectation, the present study finds that BRDSIZE is

significant and positively related to SPECLIST_MSLEADER, SPECLST_MS30 and

SPECLST_MS in the year 2005 and in the pooled models but that it is insignificantly

related to other measures. Similarly, BRDNED is found to be negatively related to

SPECLST_MS30 and SPECLST_MS in the year 2007 and in the pooled sample.

However, it is insignificant with the SPECLIST_MSLEADER, SPECLST_PS and

SPECLST_WEIGHTED. Likely, BRDEXP is also insignificant with all the other

auditor industry specialist measures. BRMEET is conditional on the auditor industry

specialist measures. It is significant and negatively related to SPECLST_MS and

positively related to SPECLST_PS, but it is not significantly related to

SPECLST_MS30, SPECLIST_MSLEADER or SPECLST_WEIGHTED.

For the SPECLIST_MSLEADER and the SPECLST_MS30 measures, the present study employs the

hetroscedastic ordinal regression and logit regression, since the dependent variable is dichotomous. The

hetroscedastic ordinal regression is estimated using the function of the oglm command in Stata.

Table 5.21: Univariate tests

SPEC_AUD SPECLST_MSLEADER SPECLST_30MS

Mean Std. Dev.

t-test

Mean Std. Dev.

t-test Specialization

(N=274)

(N=400)

(N=274)

(N=400)

(N=364)

(N=310)

(N=364)

(N=310)

BRDSIZE 9.843 8.763 2.586 1.965 -6.156*** 9.676 8.645 2.529 1.849 -5.947***

BRDNED 0.462 0.453 0.121 0.113 -0.951 0.463 0.449 0.119 0.112 -1.492

BRDEXP 0.327 0.369 0.130 0.148 3.781*** 0.341 0.364 0.136 0.149 2.039**

BRDMEET 8.504 8.730 2.400 2.703 1.117 8.574 8.713 2.357 2.830 0.694

ACSIZE 3.762 3.548 0.929 0.758 -3.301*** 3.736 3.516 0.934 0.691 -3.427***

ACIND 0.737 0.713 0.441 0.453 -0.703 0.755 0.684 0.430 0.466 -2.073**

ACEXP 0.374 0.397 0.183 0.226 1.346 0.394 0.379 0.189 0.232 -0.927

ACMEET 4.015 4.005 1.155 1.253 -0.101 3.964 4.061 1.087 1.346 1.035

INOWN 3.563 4.638 10.050 13.503 1.122 3.034 5.572 9.005 15.051 2.699***

LNASSET 6.306 6.084 0.612 0.516 -5.068*** 6.288 6.041 0.602 0.492 -5.777***

LEVERGN 0.654 0.631 0.224 0.211 -1.348 0.660 0.617 0.212 0.220 -2.586***

NWFUNDRATIO 0.108 0.149 0.261 0.542 1.181 0.102 0.168 0.235 0.611 1.872

ROA 9.949 9.604 7.764 9.946 -0.482 9.639 9.867 7.781 10.484 0.323

SQSUB 4.916 4.556 2.227 1.956 -2.218** 4.937 4.428 2.168 1.931 -3.189

FORGSALE 48.198 43.872 35.268 36.187 -1.540 49.600 40.971 35.472 35.794 -3.134***

Specialization is according to dummy definitions. The value 1 is denotes firms that employ industry specialist auditor, 0 if otherwise. The significant

level is based on two tailed tests, *** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10.

ACSIZE is significant and negatively related only to SPECLST_PS and

SPECLST_WEIGHTED in the pooled sample, while ACIND is significant and

positively related across all the auditor industry specialist measures in the pooled

samples, except in SPECLIST_MSLEADER model. This positive relationship may

suggest that the audit committees comprised solely of independent members are more

likely to appoint industry specialist auditors than non-specialist auditors. In contrast

to the prediction and the prior documented evidence, the present study finds that in

most years and in the pooled sample, ACMEET is significant and negatively related to

all the auditors industry specialist measures (except SPECLST_PS), suggesting that

firms with lower audit committee meetings frequencies are more likely to appoint

industry specialist auditors. However, ACEXP is found to be insignificantly related

with all the IND_SPEC measures.

INOWN is found to be negatively related to SPECLST_MS30 and SPECLST_MS in

pooled sample. This may indicate that, as the percentage of insider ownership

increases, there is less need for industry specialist auditors and this is due to the

volume of detailed information that is received by the directors. LNASSET and

LEVERGN suggest positive relationships with IND_SPEC in most models, suggesting

that larger firms and higher leverage firms require more specialist auditors in order to

compensate for the increases in agency costs. In contrast to the prediction,

NWFUNDRATIO is found to be negatively related to the most of SPEC_AUD

measures, while ROA, SQSUB and FORGNSALE are positively related to SPEC_AUD

in most models under the pooled samples. This indicates that higher risk firms and the

more complex firms‟ have an increased need for industry specialist auditors.

Overall, the results for board and audit committee characteristics and related control

variables are sensitive to the choice of SPEC_AUD measures. However, despite the

inconsistent results in year-by-year analysis, in the pooled model, in four out of five

of auditor industry specialist measures, ACIND and ACMEET are found to be

significantly related to the use of industry specialist auditors. The firms with audit

committees that are comprised solely of independent members are more likely to

engage industry specialist auditors. In addition, contrary to the expectation, the higher

audit committee meetings frequencies are not necessarily associated to the choice of

higher quality auditors.

Table 5.22: The results of multivariate regression for auditor industry specialist model

SPEC_AUD= α0 + β

1BRDSIZE + β

2BRDNED + β

5ACSIZE + β

6ACIND + β

7ACEXP+ β

8ACMEET + β

9INOWN +

LNASSET + β11

LEVERG + β12NWFUNDRATIO + β13ROA + β14SQSUB + β15FORGSALE + ε

The dependent variable of SPEC_AUD is measured as follows:

(1) SPECLST_MSLEADER ; (2) SPECLST_MS30; (3) SPECLST_MS; (4) SPECLST_PS; and (5) SPECLST_WEIGHTED

Variable

Coefficient (t-statistics)a

2005 (N=167) 2006 (N=181) 2007 (N=181) 2008 (N=145) Pooled (N=674)

(1) (2) (1) (2) (1) (2) (1) (2) (1) (2) (3) (4) (5)

Intercept 6.068

(2.59)**

(2.98)***

(2.00)**

(2.16)**

(1.51)

(1.84)*

(2.13)**

(2.99)**

(3.86)***

(4.93)***

-0.187

(-1.66)*

(1.47)

-0.019

(-0.65)

BRDSIZE 0.291

(2.28)**

(2.31)**

(1.37)

(0.27)

(0.50)

(0.39)

(1.41)

(1.78)*

(2.39)**

(2.09)**

(2.84)***

-0.005

(-1.50)

(0.23)

BRDNED -0.390

(-0.18)

-1.812

(-0.85)

-0.081

(-0.04)

-1.276

(-0.60)

-3.126

(-1.53)

-5.227

(-2.37)**

(0.18)

(0.26)

-0.898

(-0.90)

-1.778

(-1.74)*

-0.215

(-2.26)**

-0.045

(-0.73)

-0.037

(-1.33)

BRDEXP -0.838

(-0.48)

(0.04)

-2.449

(-1.45)

-2.980

(-1.77)*

-0.833

(-0.55)

-0.255

(-0.17)

(0.08)

(0.40)

-1.154

(-1.43)

-0.712

(-0.88)

(0.11)

-0.018

(-0.36)

-0.030

(-1.35)

BRDMEET 0.005

(0.06)

(0.15)

-0.071

(-1.06)

-0.062

(-0.99)

(1.46)

(1.64)

-0.050

(-0.58)

-0.007

(-0.08)

-0.011

(-0.31)

(0.30)

-0.007

(-2.06)**

(2.98)***

(1.38)

ACSIZE -0.218

(-0.78)

-0.177

(-0.64)

-0.359

(-1.31)

(0.62)

(1.50)

(1.63)

(0.01)

-0.242

(-0.84)

(0.17)

(0.54)

-0.017

(-2.25)**

-0.006

(-1.80)*

ACIND 0.224

(0.50)

(0.48)

(1.18)

(1.16)

-0.141

(-0.33)

(1.39)

(1.25)

(1.70)*

(1.09)

(2.25)**

(1.95)**

(2.76)***

(2.42)**

ACEXP 0.384

(0.33)

(0.42)

(0.88)

(2.11)**

(0.26)

(1.06)

-1.569

(-1.19)

-0.265

(-0.20)

(0.08)

(1.54)

-0.021

(-0.43)

(1.08)

(1.07)

ACMEET -0.326

(-1.91)*

-0.392

(-2.33)**

-0.226

(-1.37)

-0.168

(-1.03)

-0.276

(-1.65)*

-0.438

(-2.55)**

(1.62)

-0.122

(-0.58)

-0.151

(-1.92)*

-0.291

(-3.63)***

-0.021

(-3.16)***

-0.006

(-1.05)

-0.004

(-1.66)*

INOWN 0.001

(0.08)

-0.003

(-0.23)

(0.19)

-0.011

(-0.67)

-0.018

(-1.12)

-0.034

(-1.75)*

-0.017

(-0.71)

-0.028

(-1.24)

-0.006

(-0.79)

-0.016

(-2.08)**

-0.001

(-3.05)***

(0.62)

-0.000

(-0.11)

LNASSET 0.638

(1.53)

(2.47)**

(1.94)**

(1.79)*

(0.86)

(1.11)

(1.15)

(2.08)**

(2.79)***

(3.99)***

(4.35)***

-0.004

(-0.31)

(1.84)*

LEVERGN 1.609

(1.77)*

(1.70)

(0.01)

(0.67)

(0.63)

(1.80)*

-0.478

(-0.44)

(0.70)

(1.06)

(2.19)**

(1.53)

(2.74)***

(2.92)***

NWFUNDRATIO 0.758

(0.68)

(0.31)

-0.148

(-0.29)

-0.382

(-0.73)

-0.911

(-1.21)

-1.489

(-1.76)*

-1.908

(-1.08)

-0.709

(-0.41)

-0.320

(-1.01)

-0.651

(-1.78)*

-0.018

(-1.35)

-0.037

(-4.48)***

-0.013

(-3.25)***

ROA 0.008

(0.35)

-0.019

(-0.77)

(1.40)

(0.90)

(0.06)

(0.17)

(2.06)**

(1.95)*

(1.73)*

(1.34)

(1.83)*

(0.50)

(0.96)

SQSUB 0.002

(0.02)

-0.061

(-0.65)

(0.76)

(0.96)

(1.02)

(1.13)

-0.076

(-0.75)

-0.019

(-0.18)

(0.83)

(0.94)

(0.18)

(5.05)***

(3.80)***

FORGSALE 0.003

(0.58)

(0.27)

(0.06)

(1.25)

(0.69)

(1.69)

-0.005

(-0.73)

(1.62)

(0.52)

(2.57)***

(0.90)

(5.49)***

(3.24)***

Adj. R2/ Pseudo

R2 0.120 0.137 0.092 0.108 0.103 0.168 0.134 0.159 0.062 0.099 0.127 0.148 0.095

*** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10, a=z-statistics for hetroskedastic ordinal regression.

Several tests are performed to observe whether the main results of the auditor industry

specialist models are robust to various other model specifications. These tests include

the heteroscedasticity and multicollinearity check, different regression estimators,

alternative definitions of hypotheses variables, additional control variables, an

endogeneity test and the 2SLS regression.

5.6.3.1 Heteroscedasticity and multicollinearity checks

The results of heteroscedasticity test are presented in Table 5.23, according to the

specific auditor industry specialist measures. All models indicate significant p-values

that range from p<0.01 and p<0.10, suggesting that the models are heteroscedastic.68

Table 5.24 presents the result of VIF and tolerance values.69

None of the variables had

a VIF value of more than 10 or a tolerance value lower than 0.10. This may suggest

that the models had no indication of a multicollinearity problem.

Previously, SPEC_MSLEADER and SPECLST_MS30 were estimated using

heteroscedastic ordinal regression and the other measures were regressed using a least

square regression with robust standard errors, which is efficient in controlling for

heteroscedasticity. As the benchmark of comparison, in this section,

SPEC_MSLEADER and SPECLST_MS30 regressed using probit regression, while

other measures are estimated using OLS and GLS regressions. The results are

presented in Table 5.25. As can be seen, the results of GLS, OLS and Probit

regressions are relatively consistent with the main finding, suggesting that the main

results are robust to different regression estimators.

For dependent variable which is a continuous version (e.g. SPECLST_MS, SPECLST_PS and

SPECLST_WEIGHTED), the heteroscedasticity test was performed using the Breusch-Pagan or Cook-

Weisberg test, while for dichotomous version (e.g. SPECLST_MSLEADER and SPECLST_MS30), the

present study used the heteroskedasticity test for Logit/ Probit model that available in Stata program by

using the hetprob command.

69 This result is based on the modelling of SPECLST_WEIGHTED as a dependent variable. The other

models indicate relatively similar VIF and tolerance values.

As with the audit fees and NAS fees model, new definitions of board and audit

committee variables are introduced to see whether the results are robust to alternative

specifications of variables. Following the approach of Abbott et al. (2003) and

DeFond et al. (2005), the variables are: (1) BRDSIZE1, which is coded as 1 if the

firm‟s board size is less than the sample median, and 0 if otherwise; (2) BRDNED1,

which is coded as 1 if 60% of the firm‟s directors are independent, and 0 if otherwise;

(3) ACIND1, which is defined as the proportion of independent non-executive

directors on the audit committee; (4) ACEXP1, which is coded as 1 if the audit

committee has at least one director who is equipped with financial expertise, and 0 if

otherwise; (5) ACMEET1, which is is coded as 1 if the audit committee meeting

frequency is more than the sample median, and 0 if otherwise. The other variable

definitions remained unchanged.

The results are presented in Table 5.26. The main results suggest that, in the pooled

sample, four out of five SPEC_AUD measures indicate that audit committees with

solely independent members and lower audit committee meetings frequencies are

associated with a greater likelihood of industry specialist auditors being chosen.

However, when the new definitions are introduced to audit committee independence

variable and audit committee meeting frequency variable, none of these variables are

significant (except for ACMEET1 and SPECLST_MS30). In addition, the present

study finds consistent evidence that firms with audit committees with at least one

member equipped with financial expertise are more likely to employ industry

specialist auditor. In summary, the results for ACIND and ACMEET in the main

findings are sensitive to new definitions.

5.6.3.4 Additional control variables

In line with the audit fees and NAS fees models, several control variables are included

in the primary model to see whether the inclusion of these variables affect the results.

These variables are liquidity ratio (LIQ), and growth (GROWTH), which are predicted

to be positively related to the employment of industry specialist auditors. Prior

literature suggests that higher risk firms and higher growth firms are associated with

higher agency cost (Francis and Wilson, 1988; Firth and Smith, 1992) and that they

are thus likely to engage industry specialist auditor. The results are presented in Table

5.27. The results for hypothesis variables and the main control variables are relatively

similar to the main findings. LIQ and GROWTH are insignificant to all SPEC_AUD

measures. Overall, the main finding is unchanged and the additional control variables

are unlikely to affect the results.

In line with the approach used with the previous models, the present study performs

the Durbin-Wu-Hausman test (for continuous dependent variables) and the Wald test

(for dichotomous dependent variables) to see whether the corporate governance

variables are associated with the endogeneity problem. BRDSIZE, BDRNED,

BRDMEET, BRDEXP, ACSIZE, ACIND1, ACMEET and ACEXP are treated as

endogenous variables. The results of the Durbin-Wu-Hausman test are presented in

Table 5.28. As can be seen, most of the corporate governance variables show

insignificant F-statistics, except for BRDSIZE (in the SPECLST_MSLEADER model),

BDRMEET (for the SPECLIST_MS and the SPECLST_WEIGHTED measures) and

ACMEET (for most of SPEC_AUD measures), each of which suggest significant F-

statistics, indicating the presence of endogeneity.

Furthermore, 2SLS tests are performed on selected models that contain endogenous

variables in order to mitigate the potential bias in the main model of SPEC_AUD.70

Table 5.29 presents the results of 2SLS regressions. The results are relatively similar

to the main findings. ACIND is positively and significantly related with SPEC_AUD

(except SPECLST_MSLEADER), while ACMEET has a negative relationship to

SPEC_AUD, suggesting that the audit committees that are comprised solely of

independent members and that have a lower frequency of meetings are more likely to

appoint industry specialist audits. This is consistent with the inference that is made

from the main findings. The others hypotheses variables and control variables remain

unchanged.

For the models that have dichotomous dependent variables the 2SLS test are regressed using the

Stata command, ivprobit, while for the models that contained continuous dependent variables the

command is ivregress.

Table 5.23: Heteroscedasticity test for auditor industry specialist model

Dependent variable chi2 Prob > chi2

SPECLST_MSLEADER 27.99 0.000

SPECLST_30MS 35.97 0.000

SPECLIST_MS 5.67 0.017

SPECLIST_PS 33.63 0.000

SPECLST_WEIGHTED 17.01 0.000

Table 5.24: VIF and tolerance values for auditor industry specialist model

BRDSIZE 2.08 0.481

BRDNED 2.00 0.500

LNASSET 1.93 0.518

BRDEXP 1.92 0.522

ACEXP 1.72 0.581

ACSIZE 1.60 0.624

ACIND 1.38 0.726

ACMEET 1.28 0.780

FORGSALE 1.27 0.786

BRDMEET 1.21 0.825

SQSUB 1.18 0.845

LEVERG 1.12 0.896

INOWN 1.10 0.912

ROA 1.09 0.914

NWFUNDRATIO 1.03 0.975

Mean VIF 1.46

Table 5.25: The results of different estimators for auditor industry specialist model (N=674)

Variable

GLS regression OLS regression Probit regression

SPECLST_MS SPECLST_PS SPECLST_

WEIGHTED SPECLST_MS SPECLST_PS

SPECLST_

WEIGHTED

SPECLST_

MSLEADER SPECLST_MS30

Intercept -0.167

(-1.46)

(1.33)

-0.020

(-0.67)

-0.187

(-1.74)*

(1.34)

-0.019

(-0.58)

-2.702

(-3.89)***

-3.618

(-5.00)***

BRDSIZE 0.014

(2.92)***

-0.005

(-1.53)

(0.21)

(2.76)***

-0.005

(-1.44)

(0.25)

(2.41)**

(2.13)**

BRDNED -0.216

(-2.26)**

-0.047

(-0.76)

-0.038

(-1.36)

-0.215

(-2.25)**

-0.045

(-0.70)

-0.037

(-1.27)

-0.580

(-0.94)

-1.120

(-1.81)*

BRDEXP -0.005

(-0.07)

-0.015

(-0.30)

-0.030

(-1.35)

(0.11)

-0.018

(-0.35)

-0.030

(-1.29)

-0.715

(-1.45)

-0.440

(-0.90)

BRDMEET -0.006

(-1.83)*

(3.08)***

(1.58)

-0.007

(-2.02)**

(3.09)***

(1.47)

-0.007

(-0.32)

(0.28)

ACSIZE 0.006

(0.50)

-0.016

(-2.12)**

-0.006

(-1.73)*

(0.55)

-0.017

(-2.07)**

-0.006

(-1.72)*

(0.18)

(0.56)

ACIND 0.041

(1.87)*

(2.76)***

(2.41)**

(2.04)**

(2.65)***

(2.33)**

(1.09)

(2.30)**

ACEXP -0.009

(-0.19)

(1.16)

(1.27)

-0.021

(-0.43)

(1.08)

(1.10)

(0.08)

(1.56)

ACMEET -0.021

(-3.09)***

-0.006

(-1.20)

-0.005

(-1.82)**

-0.021

(-2.89)***

-0.006

(-1.12)

-0.004

(-1.92)*

-0.091

(-1.91)*

-0.177

(-3.68)***

INOWN -0.001

(-2.88)***

(0.69)

-0.000

(-0.01)

-0.001

(-2.11)**

(0.62)

-0.000

(-0.11)

-0.004

(-0.80)

-0.010

(-2.15)**

LNASSET 0.084

(4.11)***

-0.003

(-0.27)

(1.82)*

(4.56)***

-0.004

(-0.29)

(1.69)*

(2.80)***

(4.06)***

LEVERGN 0.052

(1.26)

(2.43)**

(2.54)**

(1.57)

(2.78)***

(2.88)***

(1.12)

(2.33)**

NWFUNDRATIO -0.019

(-1.53)

-0.036

(-4.76)***

-0.012

(-3.52)***

-0.018

(-1.03)

-0.037

(-3.09)***

-0.013

(-2.37)**

-0.182

(-1.01)

-0.370

(-1.87)*

ROA 0.002

(1.74)*

(0.61)

(1.05)

(2.06)**

(0.39)

(0.78)

(1.75)*

(1.37)

SQSUB 0.000

(0.10)

(5.47)***

(4.13)***

(0.16)

(4.85)***

(3.86)***

(0.86)

(0.94)

FORGSALE 0.000

(1.15)

(5.43)***

(3.36)***

(0.96)

(5.31)***

(3.28)***

(0.56)

(2.58)**

Adj. R2/ Pseudo

R2 0.120 0.153 0.100 0.127 0.148 0.095 0.062 0.100

*** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10, a=z-statistics for probit regression.

Table 5.26: The results of auditor industry specialist model for the alternative test variable

definitions (N=674)

Variable

(1) (2) (3) (4) (5)

Intercept 4.364

(3.27)***

(4.24)***

-0.124

(-0.97)

-0.014

(-0.17)

-0.049

(-1.40)

BRDSIZE1 -0.252

(-1.27)

-0.279

(-1.42)

-0.020

(-1.04)

(1.25)

(0.30)

BRDNED1 -0.101

(-0.52)

-0.203

(-1.07)

-0.007

(-0.37)

-0.021

(-1.65)*

-0.008

(-1.49)

BRDEXP -1.989

(-2.94)***

-0.863

(-1.32)

-0.080

(-1.29)

-0.016

(-0.38)

-0.036

(-1.90)*

BRDMEET -0.032

(-0.94)

-0.018

(-0.55)

-0.010

(-3.13)***

(3.24)***

(1.14)

ACSIZE 0.040

(0.31)

(0.32)

(0.66)

-0.019

(-2.73)***

-0.007

(-2.19)**

ACIND1 -0.276

(-0.82)

-0.124

(-0.42)

-0.030

(-1.09)

(1.22)

(0.68)

ACEXP1 1.073

(2.88)***

(3.13)***

(1.19)

(3.34)***

(2.95)***

ACMEET1 -0.118

(-0.58)

-0.426

(-2.06)**

-0.028

(-1.47)

-0.002

(-0.18)

-0.005

(-0.76)

INOWN -0.003

(-0.37)

-0.012

(-1.57)

-0.001

(-2.31)**

(0.73)

(0.18)

LNASSET 0.580

(3.16)***

(3.94)***

(4.54)***

-0.007

(-0.62)

(1.77)*

LEVERGN 0.442

(1.11)

(2.23)**

(1.62)

(2.85)***

(3.05)***

NWFUNDRATIO -0.277

(-0.90)

-0.631

(1.72)*

-0.018

(-1.33)

-0.036

(-4.46)***

-0.013

(-3.35)***

ROA 0.014

(1.48)

(1.00)

(1.52)

(0.28)

(0.69)

SQSUB 0.055

(1.30)

(0.90)

(5.14)***

(4.08)***

FORGSALE 0.001

(0.27)

(2.27)*

(0.26)

(5.81)***

(3.32)***

Adj. R2/ Pseudo

R2 0.062 0.085 0.099 0.150 0.093

(1) SPECLST_MSLEADER ; (2) SPECLST_MS30; (3) SPECLST_MS; (4) SPECLST_PS;

and (5) SPECLST_WEIGHTED

*** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10, a=z-statistics for

hetroskedastic ordinal regression.

Table 5.27: The results of auditor industry specialist model with the additional control

variables (N=674)

Variable

(1) (2) (3) (4) (5)

Intercept 4.458

(3.87)***

(4.90)***

-0.190

(-1.68)*

(1.40)

-0.020

(-0.69)

BRDSIZE 0.127

(2.40)**

(2.05)**

(2.88)***

-0.005

(-1.43)

(0.27)

BRDNED -1.043

(-1.03)

-1.882

(-1.83)*

-0.218

(-2.28)**

-0.051

(-0.84)

-0.042

(-1.46)

BRDEXP -1.147

(-1.42)

-0.724

(-0.89)

(0.12)

-0.017

(-0.33)

-0.030

(-1.31)

BRDMEET -0.011

(-0.30)

(0.28)

-0.007

(-2.05)*

(2.99)***

(1.39)

ACSIZE 0.016

(0.13)

(0.53)

(0.52)

-0.017

(-2.31)**

-0.006

(-1.86)*

ACIND 0.244

(1.13)

(2.28)**

(1.95)*

(2.77)***

(2.44)**

ACEXP 0.046

(0.09)

(1.59)

-0.023

(-0.46)

(1.03)

ACMEET -0.150

(-1.90)*

-0.288

(-3.58)***

-0.021

(-3.17)***

-0.006

(-1.07)

-0.004

(-1.65)*

INOWN -0.005

(-0.73)

-0.015

(-1.99)**

-0.001

(-3.07)***

(0.60)

-0.000

(-0.09)

LNASSET 0.589

(2.87)***

(4.02)***

(4.38)***

-0.003

(-0.23)

(1.97)**

LEVERGN 0.416

(1.03)

(2.11)**

(1.54)

(2.76)***

(2.88)***

NWFUNDRATIO -0.246

(-0.79)

-0.576

(-1.64)

-0.018

(-1.31)

-0.035

(-4.56)***

-0.012

(-3.32)***

ROA 0.017

(1.78)*

(1.39)

(1.82)*

(0.51)

(1.00)

SQSUB 0.037

(0.87)

(0.97)

(0.18)

(5.04)***

(3.80)***

FORGSALE 0.002

(0.62)

(2.62)***

(0.92)

(5.49)***

(3.32)***

LIQ -0.274

(-0.84)

-0.072

(-0.22)

-0.018

(-0.53)

(0.09)

-0.012

(-1.23)

GROWTH -0.002

(-0.54)

-0.003

(-0.75)

(0.19)

-0.024

(-1.14)

-0.000

(-0.26)

Adj. R2/ Pseudo

R2 0.064 0.100 0.127 0.149 0.098

(1) SPECLST_MSLEADER ; (2) SPECLST_MS30; (3) SPECLST_MS; (4) SPECLST_PS;

and (5) SPECLST_WEIGHTED

*** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10, a=z-statistics for

hetroskedastic ordinal regression.

Table 5.28: Endogeneity test for industry specialist model

Durbin-Wu-Hausman Test and Wald test

H0 = the residual of BRDSIZE, BDRNED, BRDMEET, BRDEXP, ACSIZE,

ACIND1, ACMEET and ACEXP are exogenous

Variable

Chi2 (1)

(1) (2) (3) (4) (5)

BRDSIZE 3.74

(p=0.053)

(p=0.108)

(p=0.105)

(p=0.625)

(p=0.105)

BRDNED 0.11

(p=0.744)

(p=0.506)

(p=0.825)

(p=0.779)

(p=0.825)

BRDEXP 0.24

(p=0.624)

(p=0.868)

(p=0.984)

(p=0.802)

(p=0.984)

BRDMEET 0.32

(p=0.574)

(p=0.421)

(p=0.093)

(p=0.245)

(p=0.093)

ACSIZE 0.03

(p=0.866)

(p=0.670)

(p=0.656)

(p=0.646)

(p=0.656)

ACIND1 1.01

(p=0.315)

(p=0.354)

(p=0.777)

(p=0.247)

(p=0.777)

ACEXP 0.41

(p=0.520)

(p=0.410)

(p=0.780)

(p=0.480)

(p=0.780)

ACMEET 10.06

(p=0.001)

(p=0.017)

(p=0.006)

(p=0.219)

(p=0.006) The dependent variable of SPEC_AUD is measured as follows:

(1) SPECLST_MSLEADER ; (2) SPECLST_MS30; (3) SPECLST_MS; (4) SPECLST_PS; and (5)

SPECLST_WEIGHTED

Table 5.29: The results of 2SLS regression for auditor industry specialist model (N=674)

Variable

SPECLST_MSLEADER SPECLST_

MS30 SPECLST_MS SPECLST_WEIGHTED

BRDSIZE ACMEET ACMEET BRDMEET ACMEET BRDMEET ACMEET

Intercept -2.612 (-3.76)*** -2.859 (-4.15)*** -3.743 (5.19)*** -0.158 (-1.40) -0.209 (-1.88)* -0.021 (-0.70) -0.023 (-0.82)

BRDSIZE 0.123 (3.09)**** 0.092 (2.89)**** 0.084 (2.51)** 0.013 (2.65)*** 0.016 (3.24)*** 0.000 (0.27) 0.001 (0.49)

BRDNED -0.320 (-0.51) -0.269 (-0.43) 0.903 (-1.45) -0.211 (-2.25)** -0.179 (-1.88)* -0.037 (-1.35) -0.030 (-1.09)

BRDEXP -0.449 (-0.87) -0.537 (-1.09) -0.321(-0.66) 0.003 (0.04) 0.028 (0.39) -0.029 (-1.34) -0.026 (1.18)

BRDMEET -0.001 (-0.06) 0.009 (0.39) 0.018 (0.81) -0.010 (-2.51)** -0.005 (-1.43) 0.002 (1.33) 0.002 (1.68)*

ACSIZE -0.029 (-0.37) -0.002 (-0.03) 0.030 (0.38) 0.007 (0.57) 0.004 (0.35) -0.006 (-1.83)* -0.007 (-1.93)*

ACIND 0.116 (0.88) 0.101 (0.77) 0.271 (2.05)** 0.043 (2.03)** 0.037 (1.73)* 0.015 (2.43)** 0.014 (2.32)**

ACEXP -0.091 (-0.28) -0.061 (-0.20) 0.421 (1.37) -0.020 (-0.42) -0.032 (-0.66) 0.016 (1.08) 0.014 (0.349)

ACMEET -0.103 (-2.15)** -0.245 (-3.70)*** -0.286 (-

4.38)*** -0.019 (-2.92)*** -0.039 (-4.07)*** -0.004 (-1.69)* -0.008 (-2.24)**

INOWN -0.004 (-0.86) -0.004 (-0.95) -0.010 (-2.24)** -0.002 (3.23)*** -0.001 (-3.20)*** -0.000 (-0.09) -0.000 (-0.19)

LNASSET 0.275 (2.12)** 0.407 (3.29)*** 0.557 (4.48)*** 0.088 (4.39)*** 0.095 (4.74)*** 0.010 (1.86)* 0.011 (2.10)**

LEVERGN 0.239 (0.97) 0.296(1.21) 0.588 (2.39)** 0.069 (1.67)* 0.066 (1.62) 0.033 (2.90)*** 0.034 (3.02)***

NWFUNDRATIO -0.189 (-1.04) -0.146 (-0.86) -0.338 (-1.72)* -0.017 (-1.31) -0.017 (-1.30) -0.013 (-

3.30)***

-0.013 (-

3.22)***

ROA 0.010 (1.69)* 0.012 (2.04)** 0.009 (1.58) 0.002 (1.81)* 0.002 (2.05)** 0.000 (0.98) 0.000 (1.14)

SQSUB 0.017 (0.64) 0.022 (0.84) 0.025 (0.93) 0.000 (0.21) 0.001 (0.16) 0.005 (3.83)*** 0.005 (3.84)***

FORGSALE 0.001 (0.47) 0.001 (0.78) 0.004 (2.74)*** 0.000 (0.74) 0.000 (1.11) 0.002 (3.26)*** 0.000 (3.37)***

Adj. R2

- - - 0.125 0.118 0.095 0.092

*** are significant at p<0.01, ** are significant at p<0.05 and *at p<0.10

5.7 Summary

This chapter reports the empirical findings on the effects of the characteristics of

boards of directors and audit committees on audit quality. The hypothesis variables

for the size of committees, the proportion of independent members, the financial

expertise and meeting frequencies of boards of directors and audit committees are

examined under three models of audit quality: audit fees, NAS fees and the use of

industry specialist auditors. Higher audit fees, lower NAS fees and the engagement of

auditor industry specialists are all associated with higher quality audits.

In the audit fees model, the multivariate regression suggests that independent boards

are positively related to audit fees. These may imply that independent boards use their

oversight function to demand extensive audit efforts from auditors, resulting in higher

audit fees. The other characteristics of boards and audit committees provide

inconsistent results with audit fees across the years and in the pooled samples. The

control variables are significant in the predicted direction, except of RECINV, which

is found to be insignificantly related to audit fees. The result for independent board is

robust to various model specifications including different estimators, clients‟ size

analysis, alternative definitions of hypotheses variables, additional control variables

and 2SLS regression.

There are four measures of auditor independence (FEE) examined in the NAS model:

LNNAF, LNTOTALFEE, FEERATIO1 and FEERATIO2. The present study finds

positive relationship between independent board and FEE, specifically the LNNAF

and LNTOTALFEES measures, in most year-by-year analysis and pooled samples.

FEERATIO1 and FEERATIO2 are insignificantly related to the independent board but

report similar positive coefficients. Prior studies suggest that the auditor independence

measured by the LNNAF and LNTOTALFEES are better than NAS fees ratios when it

comes to capturing the economic importance of the client-auditor relationship

(Ashbaugh et al., 2003; Larcker and Richardson, 2004). The positive relationship may

suggest that independent boards seem to support that higher provision of NAS are not

necessarily compromise the audit independence, but may enhance the quality of audits

due to the knowledge spillover effects. The knowledge spillover effects broaden the

auditors‟ knowledge and improved the audit judgment, which in turn results in a

higher audit quality. The other hypothesis variables provide inconsistent support to be

linked with NAS. Among the control variables, LNASSET is found to be the main

determinant for NAS in most year-by-year and pooled sample. The results for

independent non-executive director on boards are robust to various model

specifications and tests.

In the auditor industry specialist model, five measures of SPEC_AUD are employed:

SPECLST_MSLEADER, SPECLST_30MS, SPECLIST_MS, SPECLIST_PS and

SPECLST_WEIGHTED. The multivariate regressions of SPEC_AUD models suggest

that the characteristics of board and audit committee and related control variables are

sensitive to the choice of SPEC_AUD measures. However, in the pooled sample,

when four out of five SPEC_AUD measures are used, the results suggest that the firms

with audit committees that are comprised solely of independent members and also

have a lower number of meetings are more likely to engage industry specialist

auditors. Although these results are robust to different estimators and the 2SLS test,

they are sensitive to new variable definitions. When the definition of ACIND is

changed from sole independence to the proportion of independent members (i.e.

ACIND1) and the definition of ACMEET is changed from total number of meeting to

is dichotomous variable (coded as 1 if audit committee meeting is more than the

sample median, and 0 if otherwise), the results are no longer significant. The

summary of the hypothesis and findings are presented in Table 5.30. The significant

findings are based on the consistency of the results in year-by-year analysis and

pooled samples.

Table 5.30: The summary of the hypothesis and the findings – the relationship

between the corporate governance characteristics‟ and auditor quality.

Hypotheses Findings

H1: There is a negative relationship between the board’s

size and audit fees. Not supported

H2: There is a positive relationship between the board’s

size and NAS fees. Not supported

size and the engagement of industry specialist

auditor

Not supported

H4: There is a positive relationship between the

independent board and audit fees. Supported

H5: There is a negative relationship between the

independent board and NAS fees. Not supported

H6: There is a positive relationship between the

independent board and the engagement of industry

specialist auditor

Not supported

financial expertise and audit fees. Not supported

financial expertise and NAS fees. Not supported

financial expertise and the engagement of industry

specialist auditor

Not supported

meeting frequency and audit fees. Not supported

meeting frequency and NAS fees. Not supported

meeting frequency and the engagement of industry

specialist auditor

Not supported

H13: There is a positive relationship between the audit

committee’s size and audit fees. Not supported

H14: There is a negative relationship between the audit

committee’s size and NAS fees. Not supported

committee’s size and the engagement of industry

specialist auditor

Not supported

H16: There is a positive relationship between the solely

independent audit committee and audit fees. Not supported

H17: There is a negative relationship between the solely

independent audit committee and NAS fees. Not supported

H18: There is a positive relationship between the solely

independent audit committee and the engagement of

industry specialist auditor

Not supported

committee’s financial expertise and audit fees. Not supported

committee’s financial expertise and NAS fees. Not supported

committee’s financial expertise and the engagement

Not supported

committee’s meeting frequency and audit fees. Not supported

committee’s meeting frequency and NAS fees. Not supported

committee’s meeting frequency and the engagement

Not supported

CHAPTER 6

FINDINGS AND DISCUSSIONS: THE EFFECTIVENESS OF THE BOARD

OF DIRECTORS, AUDIT COMMITTEE, AUDIT QUALITY AND EARNINGS

MANAGEMENT

6.1 Introduction

This chapter presents the results for the second empirical analysis: the relationships

between the board of directors, audit committee and external auditor quality in

constraining earnings management. Consistent with the previous chapter, the audit

fees, NAS fees and industry specialist auditor variables represent the proxies for audit

quality, while the board and audit committee variables are measured in term of their

effectiveness characteristics (e.g. size, composition, financial expertise and meeting).

This chapter is organised as follows: next section presents the descriptive statistics,

the univariate test and the correlation matrix. This is then followed by the research

design, results and sensitivity analysis. The last section summarises this chapter.

6.2 Descriptive statistics and univariate test

This section reports the descriptive statistics and the results of the univariate test.

Table 6.1 presents the descriptive statistics for all related variables used to examine

the association between the board, audit committee, audit quality and earnings

management for the sample of 613 firm-year observations. The present study

highlights the descriptive statistics for earnings management, MTBV, LOSS and CFO,

since the other variables have fairly similar means and standard deviations, as

described in Chapter 5.1.

Across the three measures of DACC, the mean (median) of DACCJM, DACCMJM

and DACCROA are relatively consistent, at 0.056 (0.037), 0.056 (0.037) and 0.059

(0.038), respectively. Yet a study conducted by Furgeson et al. (2004), which used the

modified Jones model, found the mean and median absolute values of the sample

firms‟ discretionary accruals over the period 1996 to 1998 to be 0.092 and 0.073,

which is considerably higher than those documented in this study. This is probably

due to the reforms initiated by the regulatory agency in promoting best practices in

corporate behaviour. Such improvements, for example, can be seen from the Peasnell

at al.‟ (2000) study that suggest that in the post-Cadbury period, earnings

managements (e.g. income-increasing accruals to avoid losses or a decline in

earnings) are smaller in firms whose board of directors has a higher proportion of

non-executive directors. This is contrast to the pre-Cadbury period, during which

there is no evidence suggest the composition of non-executive directors on board is

associates to the earnings manipulation. They conclude that the publication of the

Cadbury Report (1992) had a material impact on board monitoring function by

helping firms to raise the standard of corporate behaviour, and especially the

monitoring roles of the non-executive.

The mean (median) of MTBV and CFO are 3.066 (2.95) and 0.135 (0.110),

respectively. Only 5.22% of the sample firms had two or more years of negative

income (LOSS). Previous UK study by Peasnell et al. (2005) report the mean

(median) of CFO in their sample was 0.116 (0.108), which is relatively consistent as

documented in this study.

Table 6.2 presents the results of the mean difference tests for all variables in the

earnings management model, according to the DACC measures. DACC found to be

below the median are considered „lower‟ earnings, whilst DACC above median are

identified as „higher‟ earnings. As can be seen in Table 6.2, the results of the t-tests

for all variables under the DACCJM and DACCMJM models are fairly consistent.

Most audit quality variables using these measures show significant mean differences

between the two groups of earnings, including LNAFEE, LNNAF, LNTOTALFEES,

SPECLST_MSLEADER, SPECLST_PS and SPECLST_WEIGHTED. However, under

the DACCROA definition, the results of the t-tests suggest that only the variables

SPECLST_PS and SPECLST_WEIGHTED indicate a significant difference between

the lower and higher groups of earnings management.

For the board of director variables (under the DACCJM and DACCMJM), the results

of the t-tests indicate that the BRDNED show a significant mean difference between

the lower and higher groups of earnings. However, across the three DACC measures,

there are no significant difference between these two groups of earnings for variables

on BDRSIZE, BRDEXP and BRDMEET.

With respect to audit committee characteristics, the results for ACSIZE are

conditional. Under the DACCJM definition, the results show that the mean audit

committee size differs between the lower and higher groups of earnings, whilst when

DACCMJM and DACCROA measures are used there is no significant difference. In

addition, the t-test results for ACIND suggest that the means differ between the lower

and higher groups of earnings across all measures of DACC. Among the control

variables, the mean of INOWN is found to differ significantly between the higher and

lower groups of earnings (under the DACCJM and DACCMJM measures). Other

variables provide no evidence that their means are differ between the two groups of

earnings.

Table 6.1: Descriptive statistics (N=613)

Variables Mean Standard

Deviation First Quartile Median Third Quartile Minimum Maximum

DACCJM 0.056 0.072 0.018 0.037 0.069 0.000 0.851

DACCMJM 0.056 0.071 0.018 0.037 0.068 0.000 0.817

DACCROA 0.059 0.072 0.016 0.038 0.074 0.000 0.583

LNAFEE 2.897 0.434 2.587 2.903 3.176 1.845 4.176

LNNAF 2.677 0.726 2.307 2.752 3.113 0.000 4.398

LNTOTALFEES 3.170 0.447 2.845 3.159 3.477 2.068 4.477

FEERATIO1 0.421 0.204 0.278 0.400 0.555 0.000 0.918

FEERATIO2 1.111 1.354 0.385 0.667 1.250 0.000 11.113

SPECLST_MSLEADER 0.401 0.491 0.000 0.000 1.000 0.000 1.000

SPECLST_MS30 0.530 0.499 0.000 1.000 1.000 0.000 1.000

SPECLST_MS 0.342 0.208 0.199 0.321 0.426 0.015 0.902

SPECLST_PS 0.180 0.150 0.049 0.159 0.257 0.003 0.537

SPECLST_WEIGHTED 0.068 0.067 0.013 0.045 0.097 0.000 0.228

BRDSIZE 9.083 2.229 7.00 9.000 10.000 5.000 19.000

BRDNED 0.450 0.116 0.375 0.444 0.538 0.053 0.778

BRDEXP 0.353 0.141 0.250 0.333 0.444 0.083 0.8333

BRDMEET 8.672 2.626 7.000 8.000 10.000 3.000 21.000

ACSIZE 3.572 0.792 3.000 3.000 4.000 3.000 8.000

Variables Mean Standard

Deviation First Quartile Median Third Quartile Minimum Maximum

ACIND 0.718 0.450 0.000 1.000 1.000 0.000 1.000

ACEXP 0.390 0.206 0.000 0.333 0.500 0.000 2.000

ACMEET 3.974 1.206 3.000 4.000 4.000 2.000 10.000

INOWN 4.488 12.713 0.088 0.252 1.459 0.002 147.525

BLOCK 23.588 15.795 12.010 21.280 33.250 0.000 76.120

MTBV 3.066 24.718 1.940 2.95 4.45 -318.53 194.68

LOSS 0.052 0.222 0.000 0.000 0.000 0.000 1.000

CFO 0.135 0.105 0.072 0.110 0.174 -0.063 1.273

LEVERGN 0.645 0.218 0.513 0.645 0.758 0.056 2.055

LNASSET 6.143 0.526 5.766 6.105 6.449 4.516 7.706

DACCJM=discretionary accrual based on Jones Model, DACCMJM=discretionary accruals based on Modified Jones model;

DACCROA=discretionary accruals by Kothari et al. (2005), including lagged ROA in the accrual regression to control for firm performance;

LNAFEE= the natural log of audit fees; LNNAF=natural log of NAS fees; LNTOTALFEES=natural log of the sum of audit and NAS fees;

FEERATIO1= the fee ratio of NAS fees to total fees; FEERATIO2= the fee ratio of NAS fees to audit fees; SPECLST_MSLEADER= coded as 1 if the

auditor earned the largest market share in each particular industry, 0 if otherwise; SPECLST_MS30= coded as 1 if the auditor‟ market share exceed 30

percent in each particular industry, 0 if otherwise; SPECLST_MS= continuous variable which equals to the respective auditor‟ market share;

SPECLST_PS= continuous variable which equals to the respective auditor‟ portfolio share; SPECLST_WEIGHTED= continuous variable which

equals to the compliment between auditor‟ market share (SPECLST_MS) and portfolio share (SPECLST_PS); BRDSIZE=the numbers of board‟

members during the year; BDRNED=the proportion of non-executive directors on board to board size; BRDEXP=the proportion of directors with

accounting experience and financial qualification to board size; BRDMEET= the number of board meetings during the year; ACSIZE= the number of

audit committee members; ACIND= coded as 1 if audit committee had solely non-executive directors; 0 otherwise; ACEXP= the proportion of audit

committee members with accounting experience and financial qualification to audit committee size; ACMEET= the number of audit committee

meetings during the year; INOWN=the cumulative percentage of total shares owned by the directors of a firm; BLOCK= the cumulative percentage

shares ownership of the blockholders who hold at least 5 percent or more of outstanding common shares and who are unaffiiliated with management;

MTBV= market to book value ratio; LOSS= coded as 1 if the firm had two or more years of negative income, 0 otherwise; CFO=cash flow from

operation scaled by lagged total asset; LEVERGN=proportion of debts to total assets; LNASSET= the natural logarithm of total assets;

Table 6.2: Univariate Tests

DACCJM DACCMJM DACCROA

Mean Std. Dev. t-test

lower higher lower higher lower higher lower higher lower higher lower higher

LNAFEE 2.857 2.936 0.435 0.429 -2.263** 2.857 2.937 0.432 0.433 -2.290** 2.878 2.915 0.455 0.411 -1.060

LNNAF 2.627 2.727 0.714 0.736 -1.720* 2.621 2.733 0.735 0.714 -1.912* 2.654 2.700 0.736 0.716 -0.776

LNTOTALFEES 3.123 3.216 0.452 0.438 -2.569** 3.127 3.212 0.447 0.444 -2.358** 3.152 3.188 0.466 0.428 -0.997

FEERATIO1 0.412 0.430 0.207 0.201 -1.131 0.415 0.427 0.211 0.197 -0.715 0.418 0.424 0.209 0.199 -0.344

FEERATIO2 1.076 1.147 1.330 1.378 -0.652 1.105 1.118 1.357 1.353 -0.115 1.133 1.089 1.437 1.267 0.408

SPECLST_

MSLEADER 0.359 0.443 0.481 0.498 -2.114** 0.366 0.436 0.483 0.497 -1.781* 0.389 0.414 0.488 0.493 -0.625

SPECLST_MS30 0.497 0.564 0.501 0.497 -1.658* 0.5000 0.560 0.501 0.497 -1.495 0.529 0.531 0.499 0.499 -0.038

SPECLST_MS 0.335 0.350 0.214 0.202 -0.872 0.338 0.347 0.213 0.203 -0.494 0.351 0.334 0.013 0.011 1.030

SPECLST_PS 0.150 0.211 0.131 0.161 -5.141*** 0.155 0.205 0.133 0.160 -4.219*** 0.158 0.203 0.130 0.164 -3.781***

SPECLST_

WEIGHTED 0.055 0.080 0.057 0.073 -4.857*** 0.057 0.078 0.058 0.072 -3.996*** 0.060 0.075 0.060 0.072 -2.851***

BRDSIZE 9.006 9.160 2.159 2.297 -0.850 9.049 9.117 2.173 2.287 -0.389 9.072 9.084 2.196 2.266 -0.125

BRDNED 0.438 0.463 0.113 0.117 -2.621*** 0.440 0.461 0.115 0.116 -2.168** 0.449 0.452 0.116 0.116 -0.399

BRDEXP 0.348 0.358 0.135 0.147 -0.934 0.348 0.358 0.135 0.147 -0.886 0.347 0.358 0.133 0.149 -0.950

BRDMEET 8.578 8.765 2.609 2.644 -0.882 8.611 8.733 2.644 2.611 -0.574 8.618 8.726 2.569 2.686 -0.512

ACSIZE 3.513 3.632 0.702 0.870 -1.861* 3.546 3.599 0.764 0.820 -0.837 3.621 3.652 0.853 0.724 1.509

ACIND 0.673 0.762 0.470 0.426 -2.456** 0.676 0.759 0.469 0.428 -2.275** 0.686 0.749 0.465 0.434 -1.732*

ACEXP 0.383 0.397 0.198 0.214 -0.853 0.382 0.399 0.197 0.215 -1.023 0.377 0.403 0.192 0.219 -1.535

ACMEET 3.938 4.010 1.214 1.198 -0.738 3.951 3.997 1.212 1.200 -0.470 3.899 4.049 1.153 1.253 -1.544

INOWN 5.621 3.359 15.300 9.347 2.209** 5.524 3.456 15.316 9.344 2.019** 5.063 3.916 14.475 10.667 1.117

BLOCK 23.687 23.489 15.962 15.652 0.1551 23.831 23.346 15.792 15.821 0.380 24.543 22.636 15.854 15.704 1.496

MTBV 2.163 3.967 30.504 17.113 -0.904 1.932 4.197 30.725 16.688 -1.134 3.028 3.105 31.438 15.389 -0.039

LOSS 0.918 0.977 0.274 0.150 3.302*** 0.918 0.977 0.274 0.150 3.302** 0.915 0.980 0.279 0.139 3.675***

CFO 0.146 0.124 0.106 0.104 2.621*** 0.147 0.123 0.106 0.104 2.774*** 0.144 0.126 0.103 0.107 2.032**

LEVERGN 0.637 0.653 0.214 0.221 -0.908 0.634 0.654 0.212 0.224 -1.055 0.647 0.644 0.204 0.231 0.185

LNASSET 6.154 6.131 0.501 0.550 0.549 6.160 6.126 0.498 0.553 0.797 6.187 6.099 0.525 0.524 2.074**

DACCJM=discretionary accrual based on Jones Model, DACCMJM=discretionary accruals based on Modified Jones model; DACCROA=discretionary accruals by Kothari et al.

(2005), including lagged ROA in the accrual regression to control for firm performance; LNAFEE= the natural log of audit fees; LNNAF=natural log of NAS fees;

LNTOTALFEES=natural log of the sum of audit and NAS fees; FEERATIO1= the fee ratio of NAS fees to total fees; FEERATIO2= the fee ratio of NAS fees to audit fees;

SPECLST_MSLEADER= coded as 1 if the auditor earned the largest market share in each particular industry, 0 if otherwise; SPECLST_MS30= coded as 1 if the auditor‟ market share

exceed 30 percent in each particular industry, 0 if otherwise; SPECLST_MS= continuous variable which equals to the respective auditor‟ market share; SPECLST_PS= continuous

variable which equals to the respective auditor‟ portfolio share; SPECLST_WEIGHTED= continuous variable which equals to the compliment between auditor‟ market share

(SPECLST_MS) and portfolio share (SPECLST_PS); BRDSIZE=the numbers of board‟ members during the year; BDRNED=the proportion of non-executive directors on board to

board size; BRDEXP=the proportion of directors with accounting experience and financial qualification to board size; BRDMEET= the number of board meetings during the year;

ACSIZE= the number of audit committee members; ACIND= coded as 1 if audit committee had solely non-executive directors; 0 otherwise; ACEXP= the proportion of audit

committee members with accounting experience and financial qualification to audit committee size; ACMEET= the number of audit committee meetings during the year; INOWN=the

cumulative percentage of total shares owned by the directors of a firm; BLOCK= the cumulative percentage shares ownership of the blockholders who hold at least 5 percent or more

of outstanding common shares and who are unaffiiliated with management; MTBV= market to book value ratio; LOSS= coded as 1 if the firm had two or more years of negative

income, 0 otherwise; CFO=cash flow from operation scaled by lagged total asset; LEVERGN=proportion of debts to total assets; LNASSET= the natural logarithm of total assets; The

sample populations for „lower‟ is 306 firm-year observation, while for „higher‟ is 307 firm-year observation; The significant level is based on two tailed tests, *** are significant at

p<0.01, ** are significant at p<0.05 and *at p<0.10.

Table 6.3: Pairwise correlation matrix (N=613)

A DACCJM 1.000

B DACCMJM 0.986 1.000

C DACCROA 0.850 0.854 1.000

D LNAFEE -0.049 -0.042 -0.025 1.000

E LNNAF 0.014 0.006 0.001 0.510 1.000

F LNTOTALFEES -0.028 -0.029 -0.015 0.906 0.736 1.000

G FEERATIO1 0.037 0.019 0.004 -0.113 0.662 0.300 1.000

H FEERATIO2 0.062 0.042 0.045 -0.160 0.437 0.243 0.805 1.000

I SPECLST_MSLEADER -0.054 -0.047 0.000 0.223 0.095 0.205 -0.018 -0.039 1.000

J SPECLST_MS30 -0.061 -0.049 0.001 0.279 0.139 0.246 -0.037 -0.075 0.771 1.000

K SPECLST_MS -0.017 -0.010 0.082 0.261 0.137 0.251 0.007 -0.018 0.793 0.774 1.000

L SPECLST_PS -0.195 -0.191 -0.189 0.215 0.070 0.156 -0.117 -0.113 0.423 0.431 0.187

M SPECLST_WEIGHTED -0.172 -0.167 -0.108 0.248 0.119 0.207 -0.068 -0.080 0.679 0.653 0.534

N BRDSIZE 0.092 0.097 0.094 0.423 0.270 0.423 0.043 0.016 0.243 0.204 0.260

O BRDNED -0.076 -0.080 -0.027 0.444 0.282 0.429 0.016 -0.010 0.031 0.045 0.008

P BRDEXP -0.023 -0.027 -0.019 -0.129 -0.090 -0.133 -0.022 -0.010 -0.179 -0.105 -0.137

Q BRDMEET -0.009 -0.018 -0.015 -0.017 -0.017 -0.001 0.027 0.040 -0.031 -0.008 -0.095

R ACSIZE -0.035 -0.036 0.030 0.361 0.161 0.329 -0.034 -0.063 0.102 0.107 0.099

S ACIND -0.090 -0.095 -0.057 0.132 0.109 0.125 0.003 -0.018 0.033 0.092 0.039

T ACEXP 0.035 0.039 0.010 -0.050 -0.029 -0.042 0.007 0.036 -0.051 0.021 -0.042

U ACMEET 0.029 0.025 0.007 0.340 0.264 0.365 0.092 0.079 0.007 -0.029 -0.007

V INOWN 0.029 0.026 0.023 -0.161 -0.023 -0.112 0.078 0.139 -0.040 -0.098 -0.078

W BLOCK 0.083 0.086 0.055 -0.142 -0.035 -0.093 0.075 0.134 -0.032 -0.067 -0.020

X MTBV 0.042 0.046 0.003 -0.007 0.030 0.032 0.060 0.111 0.022 0.005 0.021

Y LOSS 0.231 0.247 0.317 0.032 0.001 0.042 0.006 0.019 0.058 0.044 0.058

Z CFO 0.103 0.102 0.042 -0.179 -0.069 -0.168 0.008 0.029 -0.010 -0.050 0.011

AA LEVERGN -0.062 -0.054 -0.035 0.211 0.083 0.178 -0.061 -0.054 0.057 0.114 0.083

AB LNASSET 0.032 0.031 0.082 0.700 0.415 0.686 0.044 -0.021 0.209 0.216 0.251

L M N O P Q R S T U V

L SPECLST_PS 1.000

M SPECLST_WEIGHTED 0.882 1.000

N BRDSIZE -0.047 0.094 1.000

O BRDNED 0.057 0.034 0.013 1.000

P BRDEXP 0.083 -0.030 -0.343 0.019 1.000

Q BRDMEET 0.098 0.034 -0.134 0.146 0.018 1.000

R ACSIZE -0.062 -0.002 0.398 0.339 -0.190 0.013 1.000

S ACIND 0.139 0.103 -0.014 0.480 0.085 0.130 0.005 1.000

T ACEXP 0.094 0.044 -0.029 -0.027 0.564 -0.040 -0.204 0.076 1.000

U ACMEET 0.004 -0.005 0.242 0.247 -0.018 0.228 0.156 0.047 -0.020 1.000

V INOWN -0.022 -0.039 -0.013 -0.158 -0.022 -0.205 -0.087 -0.097 0.008 -0.132 1.000

Variable L M N O P Q R S T U V

W BLOCK -0.129 -0.109 0.046 -0.077 0.006 -0.083 -0.086 -0.079 0.065 0.065 0.057

X MTBV 0.037 0.054 0.029 0.004 0.027 -0.013 -0.040 0.027 0.001 0.059 0.010

Y LOSS 0.054 0.044 0.028 0.045 0.037 0.088 -0.025 -0.049 0.004 0.084 0.054

Z CFO -0.032 -0.036 0.019 -0.186 -0.036 -0.060 -0.072 -0.080 0.026 -0.023 0.069

AA LEVERGN 0.099 0.114 0.115 0.098 0.027 0.172 0.061 0.093 0.015 0.119 -0.182

AB LNASSET -0.039 0.083 0.443 0.398 -0.192 0.045 0.361 0.066 -0.092 0.356 -0.146

W X Y Z AA AB

W BLOCK 1.000

X MTBV 0.003 1.000

Y LOSS 0.051 0.070 1.000

Z CFO 0.110 0.088 0.159 1.000

AA LEVERGN -0.070 -0.012 -0.120 0.078 1.000

AB LNASSET -0.189 0.011 -0.045 -0.254 0.063 1.000

DACCJM=discretionary accrual based on Jones Model, DACCMJM=discretionary accruals based on Modified Jones model; DACCROA=discretionary accruals by Kothari et al.

(2005), including lagged ROA in the accrual regression to control for firm performance; LNAFEE= the natural log of audit fees; LNNAF=natural log of NAS fees;

LNTOTALFEES=natural log of the sum of audit and NAS fees; FEERATIO1= the fee ratio of NAS fees to total fees; FEERATIO2= the fee ratio of NAS fees to audit fees;

SPECLST_MSLEADER= coded as 1 if the auditor earned the largest market share in each particular industry, 0 if otherwise; SPECLST_MS30= coded as 1 if the auditor‟ market

share exceed 30 percent in each particular industry, 0 if otherwise; SPECLST_MS= continuous variable which equals to the respective auditor‟ market share; SPECLST_PS=

continuous variable which equals to the respective auditor‟ portfolio share; SPECLST_WEIGHTED= continuous variable which equals to the compliment between auditor‟ market

share (SPECLST_MS) and portfolio share (SPECLST_PS); BRDSIZE=the numbers of board‟ members during the year; BDRNED=the proportion of non-executive directors on

board to board size; BRDEXP=the proportion of directors with accounting experience and financial qualification to board size; BRDMEET= the number of board meetings during the

year; ACSIZE= the number of audit committee members; ACIND= coded as 1 if audit committee had solely non-executive directors; 0 otherwise; ACEXP= the proportion of audit

committee members with accounting experience and financial qualification to audit committee size; ACMEET= the number of audit committee meetings during the year;

INOWN=the cumulative percentage of total shares owned by the directors of a firm; BLOCK= the cumulative percentage shares ownership of the blockholders who hold at least 5

percent or more of outstanding common shares and who are unaffiiliated with management; MTBV= market to book value ratio; LOSS= coded as 1 if the firm had two or more years

of negative income, 0 otherwise; CFO=cash flow from operation scaled by lagged total asset; LEVERGN=proportion of debts to total assets; LNASSET= the natural logarithm of

total assets; Correlation in bold are significant at p<0.01, in italic are significant at p<0.05 and underline at p<0.10.

6.3 Correlation matrix

The correlation matrix for all variables indicate in the earning management model is

presented in Table 6.3. The higher correlations among discretionary accruals and audit

quality measures are always expected since they are highly interrelated to each other.

Among the audit quality proxies, the SPECLST_PS and SPECLST_WEIGHTED

variables are significant and negatively correlated with all DACC measures at p<0.01,

suggesting that the industry specialist auditors are effective in constraining

opportunistic earnings. The range of the correlation coefficient is from -0.0167 to -

0.195. None of the audit fees and NAS measures are significantly correlated with

DACC However, with respect to the independence of the board of directors and audit

committee, BRDNED and ACIND are significant and negatively correlated with

DACCJM and DACCMJM (the correlation coefficients range from -0.035 to -0.080).

Even though BRDNED and ACIND are insignificantly correlated with DACCROA,

their signs of coefficient do indicate a negative relationship. This may suggest that the

independent non-executive directors either on the board or the audit committee

contribute to the oversight of the firm and are thus likely to constrain opportunistic

earnings. The other board and audit committee characteristics are insignificantly

correlated to all DACC measures.

6.4 Multivariate regression

The multivariate regressions are estimated using least square regression with robust

standard error to control for heteroscedasticity. Table 6.4 presents the results based on

the three measures of discretionary accruals: DACCJM, DACCMJM and

DACCROA.71

Since there is multiple variables surrogate for audit quality proxy and

most of them are highly correlated with each other, each of them is included in a

single empirical model. In total there are thirty models of earnings management

examined. The F-statistic for all models is significant at p<0.01. The adjusted R2

range is between 0.108 and 0.159, which is lower than that documented in a prior UK

study conducted by Peasnell et al. (2005). This may be due to the sample size and

different model specification.

The multivariate regression was estimated only for the pooled sample, due to the insignificant F-

statistic that is obtained when it is modeled in the year-by-year samples.

As can be seen from Table 6.4, LNAFEE is significant and negatively related to all

DACC measures, suggesting that firms with higher audit fees are more likely to

constrain earnings management. There is possibility that the firms with higher audit

fees induce more audit effort, which in turn reduces the likelihood of opportunistic

earnings. This result is consistent with the argument set forth by Caramis and Lennox

(2008), who state that when audit hours are lower, firms report larger income-

increasing discretionary accruals.

In all models, none of NAS fees measures are significantly related with DACC. This

result is consistent with Chung and Kallapur (2003) and Ruddock et al. (2006), who

find no evidence of a relationship between NAS and earnings management. Although

there is no significant relationship, the results provide mix directional sign of the NAS

coefficients and earnings management, suggesting the measures of auditor

independent are sensitive to the research design.

The results for industry specialist auditors are conditional. The industry specialist

auditors measured by the SPECLST_PS and SPECLST_WEIGHTED are significant

and negatively related to earnings management across three measures of discretionary

accruals (DACCJM, DACCMJM and DACCROA). However, using the market share

approach to compute the variables of industry specialist auditors, it appears that none

of these measures are significantly related with DACC. Krishnan (2001) notes that the

portfolio approach is better at capturing auditors‟ efforts to differentiate their products

from those of their competitors than the market share approach. These significant

results may suggest that earnings management in firms with industry specialist

auditors is lower than firms with non-specialist auditors. This is consistent with

Krishnan (2003a), who suggests that the industry specialist auditors provide a higher

quality audit than non-specialist auditors by mitigating accruals-based earnings.

In relation to the board of director and audit committee characteristics, none of these

variables are significantly related with DACC except for ACSIZE, which is found to

be negatively related with earnings management in the DACCJM and DACCMJM

models for the pooled sample. Even though the relationship is weak (at p<0.10), these

results are consistent with those found by Yang and Krishnan (2005). As compared to

prior UK studies, the results are contradicted to Habbash (2010) and Habbash et al

(2010). The possible explanation on this contradictory may be due to the different

research design. Habbash (2010) segregates the board and audit committee variables

in two different earnings management model, while Habbash et al. (2010) do not

control the audit committee variables in their earnings management model.

For the control variables, INOWN and BLOCK suggest insignificant relationship with

DACC, and MTBV is negatively related with earnings management, particularly in the

DACCJM and DACCMJM models. This negative relationship contradicts the results

reported by Klein (2002), but is relatively consistent with Bowen et al. (2008). All of

the models indicate positive relationships between LOSS and CFO with DACC,

suggesting that the firms with negative income and higher cash flow have greater

incentive to manage reported earnings. The positive coefficient between CFO and

DACC is consistent with Frankel et al. (2002). In addition, there is a negative

relationship between LEVERGN and DACC in most of the DACCJM models (at

p<0.10), but it is insignificantly related to the other DACC measures. The negative

relationship between LEVERGN and DACC are consistent with prior studies (Klein,

2002; Bédard et al., 2004). LNASSET is significant and positively related with DACC

in most models, consistent with Becker et al. (1998) and DeFond and Park (1997).

In summary, the results from the multivariate analyses indicate that the firms

engaging the auditor industry specialist and paying higher audit fees are associated

with lower earnings management. This is consistent with prior studies (e.g. Caramis

and Lennox, 2008; Krishnan, 2003a) that suggest higher quality auditors have a

greater ability to constrain earnings manipulation through the extent of their

monitoring function, thus improving the quality of reported earnings. In addition,

there is no significant relationship between NAS and earnings management,

suggesting that joint provision of audit and NAS have no effect on opportunistic

earnings. This result is contradicted to the prior study in the UK done by Antle at al.

(2006) that suggests a negative relationship between NAS and earnings management.

One of the possible reasons on this may be due to the increased of NAS studies and

the reformation of the UK corporate governance system.

However, none of the results suggest that the board of director or audit committee

characteristics can be clearly linked with earnings management. As previously noted

by Larker and Richardson (2004), the monitoring role of auditors depends on the

strength of a firms‟ corporate governance structure, and therefore it is possible that the

auditor monitoring roles outweigh the oversight functions of boards and audit

committees, and hence contribute to the insignificant results for corporate governance

variables and earnings management.

Table 6.4: The results of multivariate regression for the earnings management model

DACC = β0 + β

1AQ + β

2BRDSIZE + β

3BRDNED + β

6ACSIZE + β

7ACIND + β

8ACEXP + β

9ACMEET +

INOWN + β11

BLOCK + β12

MTBV+ β13CFO + β14LEVERGN + β15LNASSET + ε

The dependent variable of DACC is measured as follows: (1) DACCJM ; (2) DACCMJM; and (3) DACCROA

The AQ proxies are: LNAFEE, LNNAF, LNTOTALFEES, FEERATIO1, FEERATIO2, SPECLST_MSLEADER, SPECLST_MS30, SPECLST_MS,

SPECLST_PS or SPECLST_WEIGHTED

Variable Coefficient (t-statistics)

(1) DACCJM

Model (a) (b) (c) (d) (e) (f) (g) (h) (i) (j)

Intercept 0.061

(1.41)

(1.63)

(1.52)

(1.61)

(1.49)

(1.48)

(1.57)

(1.98)

(1.64)

LNAFEE -0.019

(-2.23)**

LNNAF 0.001

(0.24)

LNTOTALFEES -0.014

(-1.53)

FEERATIO1 0.007

(0.48)

FEERATIO2 0.002

(0.70)

SPECLST_

MSLEADER

-0.009

(-1.51)

SPECLST_MS30 -0.008

(-1.28)

SPECLST_MS -0.009

(-0.55)

SPECLST_PS -0.078

(-4.70)***

SPECLST_

WEIGHTED

-0.167

4.92)***

BRDSIZE 0.004

(1.51)

(1.31)

(1.46)

(1.31)

(1.40)

(1.36)

(1.31)

(1.39)

(1.50)

BRDNED 0.005

(0.13)

-0.011

(-0.32)

(0.02)

-0.011

(-0.31)

-0.011

(-0.32)

-0.012

(-0.34)

-0.013

(-0.39)

-0.012

(-0.35)

-0.007

(-0.21)

-0.012

(-0.37)

BRDEXP 0.017

(0.61)

(0.57)

(0.58)

(0.61)

(0.44)

(0.49)

(0.56)

(0.60)

(0.42)

BRDMEET 0.000

(0.30)

(0.59)

(0.42)

(0.58)

(0.54)

(0.58)

(0.59)

(0.54)

(0.90)

(0.71)

ACSIZE -0.007

(-1.43)

-0.007

(-1.50)

-0.007

(-1.51)

-0.007

(-1.48)

-0.007

(-1.43)

-0.007

(-1.50)

-0.007

(-1.47)

-0.007

(-1.49)

-0.008

(-1.68)*

-0.008

(-1.66)*

ACIND -0.013

(-1.52)

-0.013

(-1.49)

-0.013

(-1.51)

-0.013

(-1.49)

-0.012

(-1.49)

-0.012

(-1.46)

-0.012

(-1.41)

-0.012

(-1.49)

-0.010

(1.23)

-0.010

(-1.26)

ACEXP 0.004

(0.26)

(0.27)

(0.28)

(0.26)

(0.23)

(0.30)

(0.37)

(0.26)

(0.53)

(0.49)

ACMEET 0.000

(0.02)

-0.001

(-0.22)

(0.04)

-0.001

(-0.24)

-0.001

(-0.30)

-0.001

(-0.33)

-0.001

(-0.37)

-0.000

(-0.26)

-0.001

(-0.26)

-0.001

(-0.46)

INOWN 0.000

(0.33)

(0.44)

(0.49)

(0.41)

(0.29)

(0.41)

(0.34)

(0.39)

(0.51)

(0.40)

BLOCK 0.000

(1.05)

(1.12)

(1.18)

(1.09)

(0.97)

(1.11)

(1.10)

(1.13)

(0.61)

(0.75)

MTBV -0.000

(-2.01)**

-0.000

(-1.99)**

-0.000

(-1.59)*

-0.000

(-2.03)**

-0.000

(-2.11)**

-0.000

(-1.88)*

-0.000

(-1.92)*

-0.000

(-1.94)*

-0.000

(-1.71)*

-0.000

(-1.63)

LOSS 0.083

(4.47)***

(4.44)***

(4.45)***

(4.44)***

(4.43)***

(4.37)***

(4.42)***

(4.30)***

(4.32)***

CFO 0.095

(2.47)**

(2.46)**

(2.48)**

(2.46)**

(2.47)**

(2.45)**

(2.44)**

(2.48)**

(2.34)**

(2.38)**

LEVERGN -0.022

(-1.40)

-0.028

(-1.76)*

-0.025

(-1.55)

-0.028

(-1.74)*

-0.027

(-1.72)*

-0.027

(-1.72)*

-0.026

(-1.70)*

-0.027

(-1.77)*

-0.024

(-1.54)

-0.023

(-1.50)

LNASSET 0.018

(2.92)***

(1.54)

(2.29)**

(1.66)*

(1.70)*

(1.94)*

(1.95)*

(1.84)*

(1.41)

(1.90)*

Adj. R2

0.113 0.108 0.111 0.108 0.109 0.111 0.110 0.108 0.132 0.130

(2) DACCMJM

Model (k) (l) (l) (n) (o) (p) (q) (r) (s) (t)

Intercept 0.070

(1.68)*

(1.86)*

(1.77)*

(1.89)*

(1.87)*

(1.75)*

(1.77)*

(1.84)*

(2.22)**

(1.87)*

LNAFEE -0.017

(-1.99)**

LNNAF -0.000

(-0.06)

LNTOTALFEES -0.014

(-1.63)

FEERATIO1 0.000

(0.03)

FEERATIO2 0.001

(0.43)

SPECLST_

MSLEADER

-0.008

(-1.35)

SPECLST_MS30 -0.006

(-0.99)

SPECLST_MS -0.006

(-0.40)

SPECLST_PS -0.074

(-4.65)***

SPECLST_

WEIGHTED

-0.160

(-4.83)***

BRDSIZE 0.004

(1.50)

(1.34)

(1.48)

(1.34)

(1.33)

(1.41)

(1.37)

(1.32)

(1.40)

(1.51)

BRDNED 0.004

(0.11)

-0.009

(-0.25)

(0.06)

-0.009

(-0.26)

-0.010

(-0.27)

-0.010

(-0.29)

-0.011

(-0.32)

-0.010

(-0.29)

-0.006

(-0.17)

-0.011

(-0.31)

BRDEXP 0.012

(0.47)

(0.44)

(0.45)

(0.44)

(0.46)

(0.31)

(0.36)

(0.42)

(0.46)

(0.28)

BRDMEET 0.000

(0.07)

(0.34)

(0.17)

(0.35)

(0.32)

(0.34)

(0.35)

(0.31)

(0.67)

(0.47)

ACSIZE -0.007

(-1.48)

-0.007

(-1.55)

-0.007

(-1.55)

-0.007

(-1.54)

-0.007

(-1.50)

-0.007

(-1.54)

-0.007

(-1.52)

-0.007

(-1.54)

-0.008

(-1.71)*

-0.008

(-1.69)*

ACIND -0.014

(-1.59)

-0.013

(-1.57)

-0.014

(-1.59)

-0.013

(-1.57)

-0.013

(-1.56)

-0.013

(-1.54)

-0.013

(-1.50)

-0.013

(-1.58)

-0.011

(-1.32)

-0.011

(-1.35)

ACEXP 0.007

(0.46)

(0.47)

(0.48)

(0.47)

(0.45)

(0.50)

(0.55)

(0.47)

(0.73)

(0.69)

ACMEET -0.000

(-0.11)

-0.001

(-0.29)

-0.000

(-0.05)

-0.001

(-0.31)

-0.001

(-0.36)

-0.001

(-0.42)

-0.001

(-0.44)

-0.001

(-0.35)

-0.001

(-0.36)

-0.001

(-0.57)

INOWN 0.000

(0.26)

(0.38)

(0.42)

(0.37)

(0.29)

(0.33)

(0.28)

(0.32)

(0.43)

(0.32)

BLOCK 0.000

(1.07)

(1.14)

(1.19)

(1.13)

(1.04)

(1.12)

(1.11)

(1.13)

(0.62)

(0.76)

MTBV -0.000

(-2.11)**

-0.000

(-2.09)**

-0.000

(-2.00)**

-0.000

(-2.09)**

-0.000

(-2.14)**

-0.000

(-2.01)**

-0.000

(-2.05)**

-0.000

(-2.06)***

-0.000

(-1.82)*

-0.000

(-1.76)*

LOSS 0.087

(4.45)***

(4.43)***

(4.44)***

(4.43)***

(4.42)***

(4.36)***

(4.37)***

(4.42)***

(4.30)***

(4.31)***

CFO 0.094

(2.57)**

(2.58)**

(2.57)**

(2.56)**

(2.54)**

(2.59)**

(2.45)**

(2.48)**

LEVERGN -0.020

(-1.28)

-0.025

(-1.60)

-0.022

(-1.39)

-0.025

(-1.61)

-0.025

(-1.58)

-0.024

(-1.57)

-0.024

(-1.56)

-0.024

(-1.62)

-0.021

(-1.38)

-0.020

(-1.34)

LNASSET 0.017

(2.76)***

(1.65)*

(2.45)**

(1.69)*

(1.72)*

(1.94)*

(1.92)*

(1.84)*

(1.44)

(1.92)*

Adj. R2

0.121 0.117 0.120 0.117 0.117 0.119 0.118 0.117 0.140 0.138

(3) DACCROA

Model (u) (v) (w) (x) (y) (z) (aa) (ab ) (ac) (ad)

Intercept 0.050

(1.01)

(1.23)

(1.13)

(1.32)

(1.27)

(1.29)

(1.30)

(1.50)

(1.59)

(1.29)

LNAFEE -0.025

(-2.54)**

LNNAF -0.003

(-0.76)

LNTOTALFEES -0.022

(-1.28)

FEERATIO1 -0.004

(-0.26)

FEERATIO2 0.002

(0.55)

SPECLST_

MSLEADER

-0.000

(-0.09)

SPECLST_MS30 -0.001

(-0.11)

SPECLST_MS 0.028

(1.56)

SPECLST_PS -0.074

(-4.63)***

SPECLST_

WEIGHTED

-0.100

2.64)***

BRDSIZE 0.003

(1.37)

(1.11)

(1.35)

(1.07)

(1.06)

(1.05)

(0.89)

(1.13)

(1.18)

BRDNED 0.011

(0.28)

-0.006

(-0.15)

(0.23)

-0.008

(-0.22)

-0.009

(-0.24)

-0.009

(-0.23)

-0.008

(-0.22)

-0.003

(-0.09)

-0.005

(-0.14)

-0.098

(-0.26)

BRDEXP 0.031

(1.24)

(1.18)

(1.20)

(1.18)

(1.22)

(1.19)

(1.20)

(1.28)

(1.22)

(1.08)

BRDMEET -0.000

(-0.22)

(0.11)

-0.000

(-0.08)

(0.17)

(0.11)

(0.16)

(0.32)

(0.45)

(0.22)

ACSIZE -0.001

(-0.13)

-0.001

(-0.24)

-0.001

(-0.21)

-0.001

(-0.22)

-0.001

(-0.18)

-0.001

(-0.21)

-0.001

(-0.21)

-0.001

(-0.19)

-0.002

(-0.36)

-0.001

(-0.29)

ACIND -0.009

(-1.15)

-0.009

(-1.10)

-0.009

(-1.14)

-0.009

(-1.11)

-0.009

(-1.10)

-0.009

(-1.11)

-0.009

(-1.11)

-0.010

(-1.26)

-0.007

(-0.83)

-0.008

(-0.95)

ACEXP -0.005

(-0.30)

-0.005

(-0.31)

-0.005

(-0.28)

-0.005

(-0.30)

-0.005

(-0.33)

-0.005

(-0.30)

-0.005

(-0.31)

-0.005

(-0.30)

-0.001

(-0.05)

-0.003

(-0.16)

ACMEET -0.003

(-1.14)

-0.003

(-1.32)

-0.003

(-0.03)

-0.003

(-1.39)

-0.004

(-1.47)

-0.004

(-1.40)

-0.004

(-1.37)

-0.003

(-1.19)

-0.004

(-1.48)

-0.004

(-1.58)

INOWN 0.000

(0.59)

(0.79)

(0.82)

(0.78)

(0.62)

(0.74)

(0.75)

(0.90)

(0.82)

(0.72)

BLOCK 0.000

(0.64)

(0.77)

(0.83)

(0.77)

(0.63)

(0.74)

(0.75)

(0.72)

(0.21)

(0.50)

MTBV 0.000

(0.57)

(0.70)

(0.76)

(0.58)

(0.70)

(0.69)

(0.62)

(1.02)

(0.94)

LOSS 0.110

(5.74)***

(5.60)***

(5.76)***

(5.61)***

(5.59)***

(5.61)***

(5.63)***

(5.69)***

(5.53)***

(5.55)***

CFO 0.073

(2.00)**

(2.02)**

(2.01)**

(1.90)*

(1.96)*

LEVERGN -0.014

(-1.04)

-0.021

(-1.55)

-0.016

(-1.21)

-0.022

(-1.59)

-0.021

(-1.55)

-0.021

(-1.57)

-0.022

(-1.59)

-0.024

(-1.64)

-0.017

(-1.31)

-0.019

(-1.39)

LNASSET 0.027

(3.56)***

(2.51)**

(3.21)***

(2.44)**

(2.47)**

(2.48)**

(2.42)**

(2.03)**

(2.22)**

(2.56)**

Adj. R2

0.146 0.137 0.145 0.137 0.137 0.137 0.137 0.142 0.159 0.145

6.6 The additional analyses and robustness tests

This section details the additional analyses that were carried out in order to see

whether the primary findings are robust in the various model specifications. The tests

include the heteroscedasticity and multicollinearity checks, various regression

estimators, alternative definitions of the board and audit committee variables, the

endogeneity test and 2SLS regressions.

6.6.1 Heteroscedasticity and multicollinearity checks

The results of the heteroscedasticity tests are presented in Table 6.5, according to the

models indicated in Table 6.4. Based on the Breusch-Pagan or Cook-Weisberg tests,

all models indicate a significant p-value, suggesting the presence of heteroscedasticity

The VIF and tolerance values are reported in Table 6.6. All models suggest that the

VIF values are between 1.00 to 2.50, and none of the variables have a VIF value of

more than 10 or a tolerance value of lower than 0.10.72 This suggests that there is no

multicollinearity problem.

6.6.2 Different regression estimators

Due to the heteroscedasticity problem, the main analyses were regressed using the

least square regression with robust standard error. This section provides the results of

the multivariate regression analysis using GLS regression, which is also efficient in

controlling for heteroscedasticity and autocorrelation, as benchmark for comparison.

The results are presented in Table 6.7. As can be seen, the results of the GLS

regression analysis are relatively consistent with the main findings.

In addition to GLS estimator, the earnings management model is also been estimated

using fixed-effect or random-effect estimators to control for time-invariant

unobserved heterogeneity in the pooled sample. The Hausman specification test was

used to show which estimator was more appropriate. Whilst the fixed-effect estimator

always provides consistent results, it might not be the most efficient; the random-

effect estimator, on the other hand, is a more efficient estimator and therefore

Since there are 30 models of earnings management examined in the main analysis, the results of the

VIF tests are reported for the selected models. However, all models suggest a relatively similar VIF

value, which is between 1.00 to 3.00.

provides a better p-value. The null hypothesis of the Hausman test is that the preferred

estimator is the random-effects. If the chi-square is more than 0.05, reject the null

hypothesis that the random effect is preferable than fixed effect estimator. The results

of the Hausman test are presented in Table 6.8. As can be seen in Table 6.8, model

(a), (b), (c), (k), (l), (m), (u), (v) and (w) indicate chi-square more than 0.05, and thus

the random-effect estimator is more appropriate. The other models were regressed

using the fixed-effect estimator.

Table 6.9 presents the results of the preferred estimator according to the Hausman

test. The results for the variables LNAFEE, SPECLST_PS, and

SPECLST_WEIGHTED are significant and negatively related with DACC in all

models. In addition, SPECLST_MSLEADER appears to have a negative relationship

with earnings management across all models. The other variables are relatively

unchanged, except ACSIZE, INOWN, MTBV and CFO, which are found to be

contradictory to the main findings. ACSIZE is no longer significant, as indicated in

models (i) and (j), while INOWN is significant and positively related with DACC in

most of the fixed effect models. MTBV is significant and positively related with

DACC in most models, while CFO is only significant when the models are estimated

using the random-effect estimator. In summary, the audit fees and industry specialist

auditor variables appear to be reasonably consistent and robust across the different

estimators.

6.6.3 New definitions for board and audit committee variables

In order to check whether the results are robust to the new variable definitions, as in

the previous chapter, the present study provides alternative definitions for the board

and audit committee variables. Following Abbott et al. (2003a) and DeFond et al.

(2005), these variables are:

(1) BRDSIZE1 - coded as 1 if the firm‟s board size is less than the sample median,

and 0 if otherwise;

(2) BRDNED1 - coded as 1 if 60% of the firm‟s directors are independent, and 0 if

otherwise;

(3) ACIND1 - defined as the proportion of independent non-executive directors on

the audit committee;

(4) ACEXP1 - coded as 1 if the audit committee has at least one director equipped

with financial expertise, and 0 if otherwise; and

(5) ACMEET1 - coded as 1 if the audit committee meetings frequency is more than

the sample median, and 0 if otherwise.

The other variables remain unchanged. Table 6.9 presents the results on the selected

models.73

As can be seen in Table 6.10, the results for the alternative definitions are relatively

consistent with the main findings except for ACIND1, which is found to be negatively

related across all the DACC measures in all the models. Furthermore, in contrast to

the primary findings, LEVERGN is insignificant with DACCJM, while LNASSET is

significant and positively related with DACCJM in Model (i). Likewise, the

BRDSIZE1, BRDNED1, BRDMEET, ACEXP1 and ACMEET variables are all

insignificant across all DACC measures. The other results are relatively unchanged, as

documented in the primary findings.

The negative relationship between ACIND and DACC suggests that firms with whose

audit committees with a higher proportion of independent non-executive directors are

likely to be associated with reduced earnings management - this is not the case,

however, for audit committees consisting solely of independent members. This may

indicate the importance of the role of executive members and of their contribution to

an effective audit committee. Overall, the results obtained for industry specialist

auditor variables hold for the selected models.

6.6.4 Endogeneity and two-stage least squares (2SLS) regression

The main results suggest that the levels of earnings management reduced when firms

hire industry specialist auditors, this is consistent with the argument that industry

specialist auditors use their industry skills and competence to constrain opportunistic

earnings. However, it can be argued that since the non-executive directors (e.g.

outsiders) have difficulty to differentiating discretionary and non-discretionary

accruals, it is possible that firms will choose industry specialist auditor to signal the

All of the models are regressed using least square regression with robust standard error.

earnings management are constrained by the presence of a higher quality auditor and

not necessarily because of their skills and competence (Francis et al., 1999 in Becker

et al., 1998).

In addition, Caramanis and Lennox (2008) argue firms which had attention to manage

earnings would have an incentive to contract for lower auditor effort (i.e. audit fees),

which suggests a negative relationship between audit fees and earnings management.

Moreover, the prior literature suggests that corporate governance characteristics are

associated with endogeneity. Thus, taking into consideration all these possibilities, the

present study tests whether the earnings management models contain these variables

indicate in the main analysis are subject to endogeneity problem.

The Durbin-Wu-Hausman test is been performed on the selected models. The results

are presented in Table 6.11. The null hypotheses are that LNAFEE, SPECLST_PS,

SPECLST_WEIGHTED, BRDSIZE, BDRNED, BRDMEET, BRDEXP, ACSIZE,

ACIND1, ACMEET and ACEXP BDRSIZE and BRDNED are exogeneous.74

If the F-

statistic were significant, then the null hypothesis would be rejected, suggesting that

the presence of endogeneity. As can be seen, the p-values of LNAFEE (in models „a‟

and „k‟), BRDMEET (in model „u‟ and „ad‟) and ACEXP (in models „i‟, „j‟ and „ac‟)

indicate significant F-statistic, suggesting that these variables are exogenous.

In order to mitigate the potential endogeneity in the models contained these variables,

the 2SLS regressions are further regressed on the selected models. Table 6.12 presents

the results of the 2SLS regressions. As can be seen, the LNAFEE, SPECLST_PS,

SPECLST_WEIGHTED are significant and negatively related with DACC in all

models. BRDSIZE and ACEXP are positively related with DACC at p<0.10,

suggesting a weak relationship with earnings management. ACIND and LEVERGN are

no longer significant under the DACCJM and DACCMJM models, and the other

variables are relatively unchanged. Likely, the results for audit fees and industry

As with the first empirical investigation, the IV variables for corporate governance and audit fees are

the lagged values, while the IV for the industry specialist auditor is the residual value from the first

regression of SPEC_AUD, consistent with Velury et al. (2003). Similar procedures are applied to all the

IV variables to ensure that they are valid (refer footnote 55).

specialist auditors SPECLST_PS and SPECLST_WEIGHTED) are consistent with the

main findings, suggesting that the main results are robust to endogeneity.

Table 6.5: Heteroscedasticity test for earnings management model

DACC= β0 + β

1AQ + β

2BRDSIZE + β

3BRDNED + β

4BRDEXP + β5BRDMEET +

β6ACSIZE + β

7ACIND + β

8ACEXP + β

9ACMEET + β

10INOWN + β

11BLOCK +

MTBV+ β13CFO + β14LEVERGN + β15LNASSET + ε

Model Dependent

variable (DACC) Audit quality proxy (AQ) chi2(1) Prob > chi2

a DACCJM LNAFEE 217.78 0.000

b DACCJM LNNAF 221.14 0.000

c DACCJM LNTOTALFEES 205.23 0.000

d DACCJM FEERATIO1 226.24 0.000

e DACCJM FEERATIO2 230.69 0.000

f DACCJM SPECLST_MSLEADER 231.21 0.000

g DACCJM SPECLST_MS30 241.32 0.000

h DACCJM SPECLST_MS 224.57 0.000

i DACCJM SPECLST_PS 318.99 0.000

j DACCJM SPECLST_WEIGHTED 301.86 0.000

k DACCMJM LNAFEE 231.75 0.000

l DACCMJM LNNAF 236.76 0.000

m DACCMJM LNTOTALFEES 224.53 0.000

n DACCMJM FEERATIO1 237.53 0.000

o DACCMJM FEERATIO2 239.95 0.000

p DACCMJM SPECLST_MSLEADER 247.12 0.000

q DACCMJM SPECLST_MS30 251.82 0.000

r DACCMJM SPECLST_MS 240.79 0.000

s DACCMJM SPECLST_PS 328.35 0.000

t DACCMJM SPECLST_WEIGHTED 312.79 0.000

u DACCROA LNAFEE 116.72 0.000

v DACCROA LNNAF 117.12 0.000

w DACCROA LNTOTALFEES 111.62 0.000

x DACCROA FEERATIO1 118.91 0.000

y DACCROA FEERATIO2 124.08 0.000

z DACCROA SPECLST_MSLEADER 119.47 0.000

aa DACCROA SPECLST_MS30 119.50 0.000

ab DACCROA SPECLST_MS 131.70 0.000

ac DACCROA SPECLST_PS 174.19 0.000

ad DACCROA SPECLST_WEIGHTED 140.92 0.000

Table 6.6: VIF and tolerance values for earnings management model

Model (a) (b) (c) (d) (f) (h) (i) (j)

Variable VIF Tolerance VIF Tolerance VIF Tolerance VIF Tolerance VIF Tolerance VIF Tolerance VIF Tolerance VIF Tolerance

LNAFEE 2.39 0.419

LNNAF 1.29 0.774

LNTOTALFEES 2.23 0.448

FEERATIO1 1.04 0.965

SPECLST_

MSLEADER 1.11 0.901

SPECLST_MS 1.16 0.865

SPECLST_PS 1.07 0.935

SPECLST_

WEIGHTED 1.06 0.944

BRDSIZE 1.99 0.503 1.95 0.512 1.99 0.504 1.94 0.509 1.96 0.510 1.98 0.506 1.94 0.517 1.94 0.515

BRDNED 2.07 0.483 2.00 0.500 2.06 0.485 1.96 0.509 1.96 0.509 1.98 0.506 1.96 0.509 1.96 0.509

BRDEXP 1.80 0.555 1.80 0.555 1.80 0.555 1.80 0.555 1.82 0.550 1.80 0.554 1.80 0.555 1.80 0.554

BRDMEET 1.21 0.829 1.19 0.842 1.19 0.837 1.19 0.844 1.18 0.845 1.19 0.840 1.19 0.841 1.18 0.844

ACSIZE 1.52 0.656 1.53 0.655 1.52 0.657 1.53 0.655 1.52 0.657 1.52 0.657 1.53 0.656 1.52 0.656

ACIND 1.43 0.698 1.43 0.697 1.43 0.697 1.43 0.698 1.44 0.696 1.44 0.693 1.44 0.693 1.44 0.693

ACEXP 1.63 0.614 1.63 0.614 1.63 0.614 1.63 0.614 1.63 0.613 1.63 0.614 1.64 0.611 1.63 0.611

ACMEET 1.32 0.760 1.32 0.757 1.33 0.753 1.31 0.763 1.31 0.762 1.32 0.760 1.30 0.767 1.31 0.765

INOWN 1.10 0.907 1.10 0.906 1.10 0.909 1.11 0.903 1.10 0.908 1.11 0.905 1.10 0.909 1.10 0.909

BLOCK 1.10 0.908 1.10 0.908 1.10 0.909 1.11 0.904 1.10 0.909 1.10 0.909 1.12 0.895 1.11 0.901

MTBV 1.02 0.976 1.02 0.977 1.02 0.977 1.03 0.975 1.02 0.977 1.02 0.977 1.02 0.976 1.03 0.975

LOSS 1.06 0.939 1.07 0.939 1.06 0.940 1.06 0.940 1.07 0.935 1.07 0.936 1.07 0.936 1.07 0.937

CFO 1.16 0.865 1.16 0.865 1.06 0.940 1.16 0.865 1.16 0.865 1.16 0.863 1.16 0.865 1.16 0.865

LEVERGN 1.16 0.863 1.11 0.899 1.14 0.878 1.11 0.898 1.11 0.899 1.12 0.896 1.12 0.896 1.12 0.894

LNASSET 2.48 0.404 1.97 0.507 2.42 0.413 1.86 0.537 1.89 0.529 1.94 0.516 1.87 0.536 1.86 0.537

Mean VIF 1.53 1.42 1.51 1.39 1.40 1.41 1.39 1.39

Table 6.7: The results of GLS regression for the earnings management model (N=613)

(1) DACCJM

Intercept -0.026

(-0.66)

-0.017

(-0.42)

-0.021

(-0.54)

-0.019

(-0.50)

-0.019

(-0.49)

-0.022

(-0.56)

-0.022

(-0.57)

-0.019

(-0.49)

(0.02)

-0.014

(-0.38)

LNAFEE -0.019

(-2.10)**

LNNAF 0.001

(0.29)

LNTOTALFEES -0.012

(-1.36)

FEERATIO1 0.007

(0.52)

FEERATIO2 0.003

(0.80)

SPECLST_

MSLEADER

-0.009

(-1.51)

SPECLST_MS30 -0.008

(-1.26)

SPECLST_MS -0.007

(-0.46)

SPECLST_PS -0.078

(-4.65)***

SPECLST_

WEIGHTED

-0.167

(-4.89)***

BRDSIZE 0.004

(1.42)

(1.22)

(1.36)

(1.23)

(1.32)

(1.28)

(1.23)

(1.31)

(1.42)

BRDNED -0.000

(-0.00)

-0.016

(-0.45)

-0.005

(-0.13)

-0.015

(-0.44)

-0.016

(-0.45)

-0.016

(-0.47)

-0.018

(-0.52)

-0.016

(-0.48)

-0.011

(-0.33)

-0.017

(-0.49)

BRDEXP 0.013

(0.49)

(0.47)

(0.51)

(0.33)

(0.38)

(0.45)

(0.50)

(0.31)

BRDMEET 0.000

(0.38)

(0.65)

(0.50)

(0.63)

(0.59)

(0.64)

(0.61)

(0.97)

(0.79)

ACSIZE -0.007

(-1.51)

-0.007

(-1.56)

-0.007

(-1.58)

-0.007

(-1.54)

-0.007

(-1.49)

-0.007

(-1.58)

-0.007

(-1.55)

-0.007

(-1.57)

-0.008

(-1.74)*

-0.008

(-1.73)*

ACIND -0.013

(-1.47)

-0.013

(-1.45)

-0.013

(-1.46)

-0.013

(-1.45)

-0.012

(-1.44)

-0.012

(-1.41)

-0.012

(-1.37)

-0.012

(-1.45)

-0.010

(-1.18)

-0.010

(-1.22)

ACEXP 0.004

(0.27)

(0.28)

(0.27)

(0.22)

(0.32)

(0.38)

(0.27)

(0.55)

(0.54)

ACMEET -0.000

(-0.06)

-0.001

(-0.31)

-0.000

(-0.06)

-0.001

(-0.33)

-0.001

(-0.41)

-0.001

(-0.40)

-0.001

(-0.45)

-0.000

(-0.33)

-0.001

(-0.35)

-0.001

(-0.55)

INOWN 0.000

(0.37)

(0.48)

(0.52)

(0.44)

(0.29)

(0.45)

(0.38)

(0.44)

(0.55)

(0.45)

BLOCK 0.000

(1.06)

(1.13)

(1.18)

(1.09)

(0.96)

(1.12)

(1.10)

(1.13)

(0.60)

(0.75)

MTBV -0.000

(-2.00)**

-0.000

(-1.99)**

-0.000

(-1.89)*

-0.000

(-2.04)**

-0.000

(-2.14)**

-0.000

(-1.87)*

-0.000

(-1.91)*

-0.000

(-1.94)*

-0.000

(-1.72)*

-0.000

(-1.64)

LOSS 0.081

(4.43)***

(4.41)***

(4.40)***

(4.33)***

(4.32)***

(4.39)***

(4.25)***

(4.28)***

CFO 0.085

(2.13)**

(2.12)**

(2.14)**

(2.11)**

(2.13)**

(2.11)**

(2.10)**

(2.13)**

(1.99)**

(2.03)**

LEVERGN -0.023

(-1.42)

-0.028

(-1.78)*

-0.025

(-1.59)

-0.028

(-1.75)*

-0.028

(-1.73)*

-0.027

(-1.75)*

-0.027

(-1.72)*

-0.028

(-1.79)*

-0.025

(-1.60)

-0.024

(-1.56)

LNASSET 0.020

(3.17)***

(1.80)*

(2.43)**

(1.93)*

(2.00)**

(2.21)**

(2.22)**

(2.08)**

(1.70)

(1.17)**

Adj. R2

0.108 0.103 0.105 0.103 0.105 0.106 0.105 0.103 0.127 0.125

(2) DACCMJM

Model (k) (l) (m) (n) (o) (p) (q) (r) (s) (t)

Intercept -0.021

(-0.56)

-0.013

(-0.36)

-0.017

(-0.47)

-0.014

(-0.37)

-0.014

(0.39)

-0.017

(-0.47)

-0.017

(-0.47)

-0.015

(-0.39)

(0.11)

-0.011

(-0.29)

LNAFEE -0.016

(-1.86)*

LNNAF -0.000

(-0.03)

LNTOTALFEES -0.013

(-1.48)

FEERATIO1 0.001

(0.08)

FEERATIO2 0.001

(0.55)

SPECLST_

MSLEADER

-0.008

(-1.35)

SPECLST_MS30 -0.006

(-0.97)

SPECLST_MS -0.005

(-0.31)

SPECLST_PS -0.074

(-4.62)***

SPECLST_

WEIGHTED

-0.160

4.82)***

BRDSIZE 0.004

(1.40)

(1.25)

(1.37)

(1.24)

(1.32)

(1.27)

(1.23)

(1.32)

(1.42)

BRDNED -0.001

(-0.04)

-0.014

(-0.38)

-0.004

(-0.10)

-0.014

(-0.39)

-0.015

(-0.40)

-0.015

(-0.43)

-0.016

(-0.45)

-0.015

(-0.42)

-0.011

(-0.30)

-0.016

(-0.45)

BRDEXP 0.009

(0.34)

(0.32)

(0.35)

(0.19)

(0.25)

(0.31)

(0.35)

(0.16)

BRDMEET 0.000

(0.16)

(0.40)

(0.26)

(0.41)

(0.37)

(0.41)

(0.39)

(0.75)

(0.56)

ACSIZE -0.007

(-1.55)

-0.007

(-1.61)

-0.007

(-1.61)

-0.007

(-1.60)

-0.007

(-1.56)

-0.007

(-1.61)

-0.007

(-1.59)

-0.007

(-1.60)

-0.008

(-1.77)*

-0.008

(-1.75)*

ACIND -0.013

(-1.54)

-0.013

(-1.53)

-0.013

(-1.54)

-0.013

(-1.53)

-0.013

(-1.52)

-0.013

(-1.50)

-0.013

(-1.46)

-0.013

(-1.54)

-0.011

(-1.27)

-0.011

(-1.31)

ACEXP 0.007

(0.49)

(0.48)

(0.50)

(0.48)

(0.46)

(0.53)

(0.57)

(0.48)

(0.76)

(0.74)

ACMEET -0.000

(-0.19)

-0.001

(-0.38)

-0.000

(-0.14)

-0.001

(-0.39)

-0.001

(-0.46)

-0.001

(-0.49)

-0.001

(-0.51)

-0.001

(-0.42)

-0.001

(-0.45)

-0.001

(-0.65)

INOWN 0.000

(0.28)

(0.39)

(0.42)

(0.39)

(0.27)

(0.36)

(0.30)

(0.35)

(0.45)

(0.35)

BLOCK 0.000

(1.06)

(1.13)

(1.17)

(1.12)

(1.02)

(1.11)

(1.10)

(1.12)

(0.60)

(0.74)

MTBV -0.000

(-2.11)**

-0.000

(-2.10)**

-0.000

(-2.01)**

-0.000

(-2.11)**

-0.000

(-2.18)**

-0.000

(-2.00)**

-0.000

(-2.05)**

-0.000

(-2.06)**

-0.000

(-1.83)*

-0.000

(-1.77)*

LOSS 0.085

(4.42)***

(4.40)***

(4.41)***

(4.40)***

(4.39)***

(4.33)***

(4.34)***

(4.39)***

(4.25)***

(4.27)***

CFO 0.085

(2.25)**

(2.26)**

(2.25)**

(2.23)**

(2.22)**

(2.26)**

(2.11)**

(2.15)**

LEVERGN -0.021

(-1.33)

-0.025

(-1.64)

-0.023

(-1.44)

-0.025

(-1.64)

-0.025

(-1.62)

-0.025

(-1.62)

-0.024

(-1.60)

-0.025

(-1.66)*

-0.022

(-1.47)

-0.022

(-1.43)

LNASSET 0.019

(3.05)***

(1.98)**

(2.67)**

(2.04)**

(2.07)**

(2.28)**

(2.25)**

(2.15)**

(1.79)*

(2.26)**

Adj. R2

0.116 0.112 0.115 0.112 0.113 0.115 0.113 0.112 0.135 0.133

(3) DACCROA

Intercept -0.061

(-1.38)

-0.052

(-1.18)

-0.056

(-1.28)

-0.049

(-1.14)

-0.051

(-1.18)

-0.050

(-1.16)

-0.050

(-1.13)

-0.043

(-1.01)

-0.033

(-0.76)

-0.048

(-1.11)

LNAFEE -0.025

(-2.44)**

LNNAF -0.002

(-0.68)

LNTOTALFEES -0.021

(-1.15)

FEERATIO1 -0.002

(-0.18)

FEERATIO2 0.002

(0.65)

SPECLST_

MSLEADER

-0.001

(-0.14)

SPECLST_MS30 0.001

(0.09)

SPECLST_MS 0.027

(1.54)

SPECLST_PS -0.074

(-4.66)***

SPECLST_

WEIGHTED

-0.103

2.80)***

BRDSIZE 0.003

(1.28)

(1.03)

(1.25)

(1.00)

(0.99)

(0.98)

(0.82)

(1.07)

(1.12)

BRDNED 0.007

(0.15)

-0.011

(-0.27)

(0.08)

-0.013

(-0.34)

-0.014

(-0.36)

-0.013

(-0.34)

-0.013

(-0.33)

-0.008

(-0.21)

-0.010

(-0.26)

-0.014

(-0.37)

BRDEXP 0.027

(1.09)

(1.05)

(1.10)

(1.05)

(1.07)

(1.17)

(1.09)

(0.94)

BRDMEET -0.000

(-0.06)

(0.25)

(0.07)

(0.29)

(0.23)

(0.28)

(0.43)

(0.60)

(0.37)

ACSIZE -0.001

(-0.27)

-0.002

(-0.38)

-0.002

(-0.36)

-0.002

(-0.36)

-0.002

(-0.31)

-0.002

(-0.35)

-0.002

(-0.35)

-0.002

(-0.33)

-0.002

(-0.50)

-0.002

(-0.44)

ACIND -0.009

(-1.07)

-0.009

(-1.04)

-0.009

(-1.07)

-0.009

(-1.04)

-0.008

(-1.04)

-0.009

(-1.05)

-0.009

(-1.05)

-0.010

(-1.19)

-0.006

(-0.76)

-0.007

(-0.88)

ACEXP -0.005

(-0.31)

-0.005

(-0.34)

-0.005

(-0.30)

-0.005

(-0.33)

-0.006

(-0.37)

-0.005

(-0.33)

-0.005

(-0.34)

-0.006

(-0.35)

-0.001

(-0.06)

-0.003

(-0.16)

ACMEET -0.003

(-1.23)

-0.004

(-1.43)

-0.003

(-1.14)

-0.004

(-1.51)

-0.004

(-1.61)

-0.004

(-1.52)

-0.004

(-1.49)

-0.003

(-1.32)

-0.004

(-1.61)

-0.004

(-1.70)

INOWN 0.000

(0.65)

(0.85)

(0.87)

(0.83)

(0.65)

(0.80)

(0.81)

(0.95)

(0.88)

(0.79)

BLOCK 0.000

(0.71)

(0.84)

(0.89)

(0.83)

(0.68)

(0.81)

(0.82)

(0.79)

(0.25)

(0.54)

MTBV 0.000

(0.61)

(0.72)

(0.78)

(0.73)

(0.58)

(0.72)

(0.66)

(1.04)

(0.96)

LOSS 0.108

(5.71)***

(5.58)***

(5.72)***

(5.58)***

(5.55)***

(5.57)***

(5.59)***

(5.66)***

(5.48)***

(5.51)***

CFO 0.065

(1.78)*

(1.80)*

(1.79)*

(1.80)*

(1.67)*

(1.96)*

LEVERGN -0.015

(-1.10)

-0.021

(-1.59)

-0.017

(-1.27)

-0.022

(-1.63)

-0.021

(-1.59)

-0.022

(-1.62)

-0.022

(-1.63)

-0.024

(-1.75)*

-0.019

(-1.40)

-0.019

(-1.45)

LNASSET 0.028

(3.75)***

(2.69)***

(3.31)***

(2.65)***

(2.69)***

(2.70)***

(2.64)***

(2.26)**

(2.43)**

(2.76)**

Adj. R2

0.143 0.134 0.141 0.134 0.135 0.134 0.133 0.139 0.156 0.142

Table 6.8: Fixed or random effect estimator

Hausman test

H0 = The preferred model is random effect

Reject H0 if chi-square was more than 0.05.

Model Dependent variable (DACC) chi2(16) Prob > chi2

a DACCJM 25.49 0.061

b DACCJM 24.22 0.085

c DACCJM 23.92 0.091

d DACCJM 30.17 0.017

e DACCJM 29.58 0.020

f DACCJM 41.19 0.000

g DACCJM 26.81 0.043

h DACCJM 28.70 0.026

i DACCJM 30.04 0.017

j DACCJM 30.14 0.017

k DACCMJM 25.53 0.061

l DACCMJM 24.68 0.075

m DACCMJM 23.47 0.102

n DACCMJM 30.20 0.017

o DACCMJM 37.95 0.001

p DACCMJM 29.63 0.020

q DACCMJM 27.80 0.033

r DACCMJM 27.88 0.032

s DACCMJM 30.48 0.015

t DACCMJM 30.65 0.014

u DACCROA 26.21 0.051

v DACCROA 25.75 0.057

w DACCROA 24.13 0.086

x DACCROA 30.18 0.017

y DACCROA 38.50 0.001

z DACCROA 28.93 0.024

aa DACCROA 26.03 0.053

ab DACCROA 28.93 0.024

ac DACCROA 27.85 0.033

ad DACCROA 27.22 0.039

Table 6.9: The results of fixed-effect/ random effect estimators for the earnings management model (N=613)

Variable Coefficient (z-statistics)

(1) DACCJM

Intercept -0.031

(-0.68)

-0.022

(-0.48)

-0.028

(-0.60)

-0.709

(-3.12)***

-0.699

(-3.05)***

-0.730

(-3.33)***

-0.741

(-3.27)***

-0.732

(-3.30)***

-0.740

(-3.39)***

-0.759

(-3.48)***

LNAFEE -0.017

(-1.95)*

LNNAF 0.001

(0.34)

LNTOTALFEES -0.013

(-1.38)

FEERATIO1 -0.018

(-0.78)

FEERATIO2 -0.005

(-1.40)

SPECLST_

MSLEADER

-0.038

(-2.78)***

SPECLST_MS30 -0.007

(-0.73)

SPECLST_MS -0.070

(-1.55)

SPECLST_PS -0.176

(-3.88)***

SPECLST_

WEIGHTED

-0.350

(-5.30)***

BRDSIZE 0.004

(1.59)

(1.40)

(1.54)

(1.13)

(1.10)

(1.23)

(1.21)

(1.17)

(1.41)

(1.35)

BRDNED -0.009

(-0.25)

-0.005

(-0.15)

(0.18)

(0.68)

(0.64)

(0.44)

-0.040

(-0.67)

(0.51)

(0.40)

BRDEXP 0.019

(0.60)

(0.56)

(0.57)

(0.88)

(0.87)

(0.88)

(0.89)

(0.84)

(1.03)

(0.91)

BRDMEET 0.001

(0.44)

(0.65)

(0.53)

-0.001

(-0.31)

-0.001

(-0.28)

-0.001

(-0.45)

(0.39)

-0.001

(-0.54)

-0.001

(-0.27)

-0.001

(-0.34)

ACSIZE -0.006

(-1.26)

-0.007

(-1.30)

-0.007

(-1.30)

-0.007

(-0.92)

-0.007

(-0.97)

-0.004

(-0.60)

-0.007

(-0.94)

-0.006

(-0.83)

-0.006

(-0.80)

-0.006

(-0.82)

ACIND -0.012

(-1.34)

-0.011

(-1.29)

-0.012

(-1.32)

-0.011

(-1.30)

-0.011

(-1.27)

-0.011

(-1.26)

-0.011

(-1.33)

-0.012

(-1.35)

-0.011

(-1.23)

-0.010

(-1.11)

ACEXP 0.002

(0.15)

(0.16)

(0.17)

-0.007

(-0.26)

-0.006

(-0.24)

-0.007

(-0.28)

-0.007

(-0.27)

-0.006

(-0.23)

-0.010

(-0.38)

-0.008

(-0.31)

ACMEET -0.001

(-0.39)

-0.001

(-0.59)

-0.001

(-0.37)

-0.001

(-0.26)

-0.001

(-0.23)

-0.001

(-0.25)

-0.002

(-0.29)

-0.002

(-0.32)

-0.002

(-0.26)

-0.002

(-0.29)

INOWN 0.000

(0.64)

(0.73)

(0.79)

(2.00)**

(2.22)**

(1.89)*

(1.79)*

(1.90)*

(2.13)**

(2.14)**

BLOCK 0.000

(1.11)

(1.15)

(1.21)

(0.39)

(0.44)

(0.21)

(0.45)

(0.38)

(0.28)

(0.21)

MTBV -0.000

(-1.99)**

-0.000

(-1.97)**

-0.000

(-2.01)*

-0.000

(-1.83)*

-0.000

(-1.75)*

-0.000

(-1.74)*

-0.000

(-1.75)*

-0.000

(-1.75)*

-0.000

(-1.93)*

-0.000

(-1.99)**

LOSS 0.081

(4.07)***

(4.04)***

(4.06)***

(2.78)***

(2.88)***

(2.87)***

(2.89)***

(2.87)***

(2.93)***

(2.92)***

CFO 0.086

(2.16)**

(2.12)**

-0.085

(-2.16)**

-0.033

(-0.54)

-0.031

(-0.52)

-0.037

(-0.59)

(0.60)

-0.037

(-0.59)

-0.036

(-0.59)

-0.036

(-0.58)

LEVERGN -0.025

(-1.49)

-0.030

(-1.80)*

-0.027

(-1.60)

-0.018

(-0.43)

-0.020

(-0.47)

-0.016

(-0.36)

-0.019

(-0.44)

-0.015

(-0.34)

-0.012

(-0.29)

-0.010

(-0.22)

LNASSET 0.019

(2.78)***

(1.56)

(2.19)***

(2.98)***

(2.90)**

(3.15)***

(3.06)***

(3.19)***

(3.24)***

(3.31)***

0.112 0.107 0.074 0.121 0.122 0.126 0.122 0.125 0.137 0.142

(2) DACCMJM

Model (k) (l) (m) (n) (o) (p) (q) (r) (s) (t)

Intercept -0.021

(-0.48)

-0.014

(-0.31)

-0.017

(-0.47)

-0.699

(-3.11)***

-0.392

(-3.05)***

-0.722

(-3.31)***

-0.724

(-3.23)***

-0.723

(-3.28)***

-0.732

(-3.37)***

-0.750

3.45)***

LNAFEE -0.016

(-1.85)*

LNNAF -0.000

(-0.02)

LNTOTALFEES -0.013

(-1.48)

FEERATIO1 -0.021

(-0.91)

FEERATIO2 -0.005

(-1.42)

SPECLST_

MSLEADER

-0.033

(-2.34)**

SPECLST_MS30 -0.001

(-0.12)

SPECLST_MS -0.030

(-0.61)

SPECLST_PS -0.163

(-2.85)***

SPECLST_

WEIGHTED

-0.329

4.22)***

BRDSIZE 0.004

(1.59)

(1.43)

(1.37)

(0.81)

(0.77)

(0.88)

(0.86)

(1.06)

(1.01)

BRDNED 0.006

(0.17)

-0.005

(-0.16)

-0.004

(-0.10)

(0.82)

(0.79)

(0.62)

(0.83)

(0.76)

(0.56)

(0.55)

BRDEXP 0.013

(0.46)

(0.43)

(0.32)

(0.82)

(0.81)

(0.82)

(0.80)

(0.95)

(0.84)

BRDMEET 0.000

(0.14)

(0.36)

(0.26)

-0.001

(-0.44)

-0.001

(-0.42)

-0.001

(-0.57)

-0.001

(-0.52)

-0.001

(-0.58)

-0.001

(-0.42)

-0.001

(-0.48)

ACSIZE -0.007

(-1.34)

-0.007

(-1.39)

-0.007

(-1.61)

-0.006

(-0.87)

-0.007

(-0.93)

-0.004

(-0.58)

-0.006

(-0.89)

-0.006

(-0.83)

-0.005

(-1.75)

-0.005

(-0.76)

ACIND -0.013

(-1.46)

-0.013

(-1.42)

-0.013

(-1.54)

-0.013

(-1.58)

-0.013

(-1.55)

-0.013

(-1.54)

-0.014

(-1.61)

-0.014

(-1.62)

-0.013

(-1.49)

-0.012

(-1.37)

ACEXP 0.006

(0.38)

(0.39)

(0.50)

-0.001

(-0.05)

-0.001

(-0.03)

-0.002

(-0.07)

-0.002

(-0.07)

-0.001

(-0.05)

-0.004

(-0.16)

-0.003

(-0.10)

ACMEET -0.000

(-0.37)

-0.001

(-0.54)

-0.000

(-0.14)

-0.002

(-0.32)

-0.002

(-0.29)

-0.002

(-0.32)

-0.002

(-0.33)

-0.002

(-0.35)

-0.002

(-0.33)

-0.002

(-0.35)

INOWN 0.000

(0.42)

(0.54)

(0.42)

(1.57)

(1.75)*

(1.46)

(1.35)

(1.41)

(1.69)*

(1.68)*

BLOCK 0.000

(1.08)

(1.13)

(1.17)

(0.30)

(0.36)

(0.17)

(0.39)

(0.35)

(0.22)

(0.14)

MTBV -0.000

(-1.87)*

-0.000

(-1.86)*

-0.000

(-2.01)**

-0.000

(-1.71)*

-0.000

(-1.66)*

-0.000

(-1.61)

-0.000

(-1.64)

-0.000

(-1.62)

-0.000

(-1.77)*

-0.000

(-1.83)*

LOSS 0.086

(4.28)***

(4.25)***

(4.41)***

(2.92)***

(3.02)***

(3.01)***

(3.06)***

CFO 0.089

(2.35)**

(2.32)**

(2.26)**

-0.040

(-0.66)

-0.039

(-0.66)

-0.046

(-0.72)

-0.046

(-0.72)

-0.046

(-0.72)

-0.045

(-0.72)

-0.044

(-0.72)

LEVERGN -0.022

(-1.33)

-0.026

(-1.61)

-0.023

(-1.44)

-0.016

(-0.38)

-0.018

(-0.43)

-0.014

(-0.33)

-0.017

(-0.40)

-0.015

(-0.36)

-0.010

(-0.25)

-0.008

(-0.19)

LNASSET 0.017

(2.61)***

(1.61)

(2.67)**

(2.97)***

(2.91)**

(3.13)***

(3.01)***

(3.12)***

(3.22)***

(3.28)***

Adj. R2

0.112 0.117 0.115 0.131 0.134 0.137 0.129 0.130 0.143 0.148

(3) DACCROA

Intercept -0.062

(-1.29)

-0.053

(-1.09)

-0.056

(-1.28)

-0.656

(-3.05)***

-0.660

(-3.05)***

-0.690

(-3.30)***

-0.687

(-3.21)***

-0.689

(-3.27)***

-0.695

(-3.32)***

-0.704

3.35)***

LNAFEE -0.024

(-2.56)**

LNNAF -0.003

(-0.71)

LNTOTALFEES -0.021

(-1.15)

FEERATIO1 -0.031

(-1.35)

FEERATIO2 -0.005

(-1.49)

SPECLST_

MSLEADER

-0.028

(-2.07)**

SPECLST_MS30 0.001

(0.10)

SPECLST_MS -0.003

(-0.06)

SPECLST_PS -0.107

(-1.78)*

SPECLST_

WEIGHTED

-0.176

(-1.87)*

BRDSIZE 0.003

(1.59)

(1.30)

(1.25)

(0.37)

(0.35)

(0.45)

(0.44)

(0.45)

(0.57)

(0.51)

BRDNED 0.015

(0.39)

-0.000

(-0.01)

(0.08)

(1.63)

(1.61)

(1.55)

(1.63)

(1.64)

(1.57)

(1.59)

BRDEXP 0.031

(1.15)

(1.09)

(1.05)

(1.07)

(1.08)

(1.07)

(1.17)

(1.09)

BRDMEET -0.000

(-0.06)

(0.25)

(0.07)

(0.68)

(0.65)

(0.51)

(0.55)

(0.53)

(0.63)

(0.58)

ACSIZE -0.001

(-0.15)

-0.001

(-0.25)

-0.002

(-0.36)

-0.003

(-0.40)

-0.004

(-0.46)

-0.002

(-0.20)

-0.004

(-0.43)

-0.004

(-0.42)

-0.003

(-0.34)

-0.003

(-0.36)

ACIND -0.009

(-1.12)

-0.009

(-1.05)

-0.009

(-1.07)

-0.013

(-1.31)

-0.013

(-1.31)

-0.013

(-1.35)

-0.014

(-1.41)

-0.014

(-1.41)

-0.013

(-1.33)

-0.013

(-1.30)

ACEXP -0.006

(-0.37)

-0.006

(-0.38)

-0.005

(-0.30)

-0.013

(-0.53)

-0.013

(-0.52)

-0.014

(-0.58)

-0.014

(-0.58)

-0.014

(-0.58)

-0.016

(-0.64)

-0.015

(-0.59)

ACMEET -0.003

(-1.17)

-0.003

(-1.31)

-0.003

(-1.14)

(-1.41))

(0.43)

(0.40)

(0.39)

(0.38)

INOWN 0.000

(0.82)

(1.04)

(0.87)

(2.13)**

(2.19)**

(1.77)*

(1.68)*

(1.69)*

(1.97)*

(1.92)*

BLOCK 0.000

(0.76)

(0.88)

(0.89)

(0.76)

(0.88)

(0.68)

(0.90)

(0.77)

(0.74)

MTBV 0.000

(0.48)

(0.61)

(0.78)

-0.000

(-0.39)

-0.000

(-0.18)

-0.000

(-0.35)

-0.000

(-0.39)

-0.000

(-0.39)

-0.000

(-0.30)

-0.000

(-0.26)

LOSS 0.111

(5.53)***

(5.40)***

(5.72)***

(4.89)***

(5.06)***

(5.05)***

(5.01)***

(5.08)***

CFO 0.070

(1.89)*

(1.88)*

(1.80)*

(0.20)

(0.17)

(0.07)

(0.08)

LEVERGN -0.016

(-1.10)

-0.022

(-1.56)

-0.017

(-1.27)

-0.017

(-0.41)

-0.019

(-0.47)

-0.016

(-0.38)

-0.019

(-0.44)

-0.019

(-0.44)

-0.015

(-0.34)

-0.014

(-0.33)

LNASSET 0.027

(3.40)***

(2.34)***

(3.31)***

(2.78)***

(2.77)***

(2.99)***

(2.87)***

(2.94)***

(3.04)***

(3.05)***

0.146 0.137 0.141 0.188 0.188 0.189 0.184 0.189 0.189 0.189

Table 6.10: The results of earning management model for the alternative test variable definitions (N=613)

Dependent

variable

Coefficient (z-statistics)

Model (a) (i) (j) (k) (s) (t) (u) (ac) (ad)

Intercept 0.019

(0.39)

(0.89)

(0.62)

(0.32)

(0.80)

(0.52)

-0.044

(-0.88)

-0.017

(-0.34)

-0.030

(-0.60)

LNAFEE -0.019

(-2.58)**

-0.017

(-2.29)**

-0.025

(-2.89)***

SPECLST_PS -0.078

(-4.85)***

-0.076

(-4.88)***

-0.075

(-4.83)***

SPECLST_

WEIGHTED

-0.161

(-4.95)***

-0.156

(-4.92)***

-0.095

(-2.54)**

BRDSIZE1 0.007

(0.89)

(0.79)

(0.87)

(0.66)

(0.57)

(0.65)

(0.22)

(0.07)

(0.15)

BRDNED1 0.005

(0.53)

(0.42)

(0.34)

(0.54)

(0.44)

(0.36)

(0.20)

(0.06)

(0.01)

BRDEXP 0.015

(0.60)

(0.64)

(0.39)

(0.33)

(0.37)

(0.12)

(0.78)

(0.81)

(0.62)

BRDMEET -0.000

(-0.32)

(0.28)

(0.02)

-0.001

(-0.57)

(0.01)

-0.000

(-0.26)

-0.001

(-0.74)

-0.000

(-0.08)

-0.000

(-0.35)

ACSIZE -0.005

(-1.35)

-0.007

(-1.89)**

-0.007

(-1.85)*

-0.005

(-1.41)

-0.007

(-1.91)*

-0.007

(-1.87)*

(0.44)

-0.000

(-0.05)

(0.04)

ACIND1 -0.022

(-1.65)

-0.022

(-1.69)*

-0.023

(-1.73)*

-0.024

(-1.71)*

-0.023

(-1.76)*

-0.024

(-1.79)*

-0.018

(-1.88)*

-0.018

(-1.92)*

-0.019

(-1.97)*

ACEXP1 -0.007

(-0.48)

-0.002

(-0.15)

-0.003

(-0.18)

-0.005

(-0.04)

(0.32)

(0.30)

-0.003

(-0.28)

(0.09)

-0.001

(-0.08)

ACMEET1 0.008

(1.14)

(1.07)

(0.99)

(0.92)

(0.85)

(0.77)

-0.001

(-0.09)

-0.001

(-0.22)

-0.002

(-0.28)

INOWN 0.000

(0.38)

(0.62)

(0.54)

(0.40)

(0.64)

(0.55)

(0.62)

(0.92)

(0.82)

BLOCK 0.000

(1.31)

(0.85)

(1.01)

(1.32)

(0.87)

(1.02)

(0.96)

(0.43)

(0.73)

MTBV -0.000

(-1.93)*

-0.000

(-1.71)*

-0.000

(-1.63)

-0.000

(-2.02)**

-0.000

(-1.78)*

-0.000

(-1.72)**

(0.56)

(0.97)

(0.87)

LOSS 0.083

(4.38)***

(4.21)***

(4.23)***

(4.34)***

(4.19)***

(4.20)***

(5.56)***

(5.37)***

(5.39)***

CFO 0.097

(2.51)**

(2.37)**

(2.41)**

(2.62)**

(2.48)**

(2.51)**

(2.01)**

(1.89)*

(1.96)*

LEVERGN -0.021

(-1.28)

-0.022

(-1.47)

-0.022

(-1.43)

-0.019

(-1.18)

-0.020

(-1.32)

-0.019

(-1.28)

-0.013

(-0.91)

-0.017

(-1.24)

-0.018

(-1.33)

LNASSET 0.020

(3.39)***

(1.80)*

(2.27)**

(3.36)*

(1.99)**

(2.46)**

(3.96)**

(2.54)**

(2.84)***

Adj. R2

0.110 0.129 0.126 0.116 0.135 0.132 0.143 0.156 0.140

Table 6.11: Endogeneity test for the earnings management model

H0 = the residual of LNAFEE, SPECLST_PS, SPECLST_WEIGHTED, BRDSIZE, BDRNED, BRDMEET, BRDEXP, ACSIZE, ACIND1, ACMEET and

ACEXP are exogenous

Variable

Chi2 (1)

Model (a) Model (i) Model (j) Model (k) Model (s) Model (t) Model (u) Model (ac) Model (ad)

LNAFEE 5.621

(p=0.017)

(p=0.015)

(p=0.116)

SPECLST_

(p=0.176)

(p=0.171)

(p=0.484)

SPECLST_

WEIGHTED

(p=0.208)

(p=0.306)

(p=0.644)

BRDSIZE 1.826

(p=0.176)

(p=0.175)

(p=0.208)

(p=0.306)

(p=0.314)

(p=0.363)

(p=0.572)

(p=0.537)

(p=0.534)

BRDNED 0.353

(p=0.552)

(p=0.458)

(p=0.428)

(p=0.629)

(p=0.541)

(p=0.509)

(p=0.266)

(p=0.181)

(p=0.169)

BRDEXP 0.842

(p=0.358)

(p=0.279)

(p=0.280)

(p=0.478)

(p=0.399)

(p=0.398)

(p=0.307)

(p=0.318)

BRDMEET 0.312

(p=0.575)

(p=0.929)

(p=0.809)

(p=0.612)

(p=0.954)

(p=0.835)

(p=0.037)

(p=0.307)

(p=0.078)

ACSIZE 0.633

(p=0.426)

(p=0.352)

(p=0.355)

(p=0.813)

(p=0.727)

(p=0.731)

(p=0.962)

(p=0.920)

(p=0.916)

ACIND1 0.860

(p=0.353)

(p=0.335)

(p=0.346)

(p=0.369)

(p=0.352)

(p=0.362)

(p=0.313)

(p=0.299)

(p=0.315)

ACEXP 2.564

(p=0.109)

(p=0.084)

(p=0.087)

(p=0.145)

(p=0.116)

(p=0.119)

(p=0.094)

(p=0.108)

ACMEET 0.169

(p=0.680)

(p=0.552)

(p=0.495)

(p=0.641)

(p=0.518)

(p=0.463)

(p=0.165)

(p=0.114)

(p=0.113)

Table 6.12: The results of 2SLS regression for earnings management model

Dependent variable

Coefficient (z-statistics)

Model (a) (i) (j) (k) (u) (ac) (ad)

Intercept -0.035

(-0.88)

(0.06)

-0.015

(-0.38)

-0.029

(-0.78)

-0.047

(-1.06)

-0.033

(-0.74)

-0.035

(-0.80)

LNAFEE -0.047

(-3.07)***

-0.044

(-2.98)***

-0.027

(-2.80)***

SPECLST_PS -0.086

(-4.59)***

-0.082

(-4.59)***

SPECLST_

WEIGHTED

-0.185

(-4.72)***

-0.097

(-2.64)***

BRDSIZE 0.005

(1.75)*

(0.11)

(0.24)

(1.74)*

(1.17)

-0.000

(-0.16)

(0.99)

BRDNED 0.027

(0.68)

-0.016

(-0.45)

-0.022

(-0.62)

(0.62)

(0.32)

-0.014

(-0.35)

-0.010

(-0.26)

BRDEXP 0.018

(0.67)

-0.096

(-1.39)

-0.100

(-1.45)

(0.53)

(1.17)

-0.076

(-1.12)

(1.02)

BRDMEET -0.00

(-0.12)

(0.94)

(0.74)

(-0.36)

-0.001

(-1.43)

(0.50)

-0.001

(-0.85)

ACSIZE -0.006

(-1.37)

-0.001

(-0.26)

-0.001

(-0.25)

-0.006

(-1.42)

-0.000

(-0.09)

(0.67)

-0.007

(-0.26)

ACIND -0.013

(-1.57)

-0.011

(-1.29)

-0.011

(-1.32)

-0.014

(-1.64)

-0.008

(-1.04)

-0.007

(-0.90)

-0.006

(-0.86)

ACEXP 0.004

(0.25)

(1.75)*

(1.73)*

(0.45)

-0.005

(-0.31)

(1.58)

-0.003

(-0.17)

ACMEET 0.001

(0.31)

(0.16)

-0.000

(-0.06)

(0.19)

-0.002

(-0.79)

-0.003

(-1.06)

-0.003

(-1.29)

INOWN 0.000

(0.16)

(0.48)

(0.36)

(0.07)

(0.33)

(0.78)

(0.50)

BLOCK 0.000

(0.97)

(0.33)

(0.49)

(0.98)

(0.60)

-0.000

(-0.05)

(0.47)

MTBV -0.000

(-2.02)**

-0.000

(-1.12)

-0.000

(-0.99)

-0.000

(-2.12)**

(0.53)

(1.33)

(0.91)

LOSS 0.082

(4.54)***

(4.35)***

(4.37)***

(4.52)***

(5.96)***

(5.49)***

(5.73)***

CFO 0.095

(2.50)**

(2.15)**

(2.19)**

(2.60)***

(2.01)**

(1.69)*

(1.98)**

LEVERGN -0.014

(-0.96)

-0.021

(-1.33)

-0.020

(-1.29)

-0.012

(-0.84)

-0.011

(-0.81)

-0.015

(-1.06)

-0.016

(-1.23)

LNASSET 0.029

(3.58)***

(1.43)

(1.96)*

(3.58)***

(3.70)***

(2.24)**

(2.57)**

Adj. R2

0.102 0.055 0.054 0.110 0.143 0.089 0.142

6.7 Summary

This chapter presents the empirical findings regarding the relationship between the

effectiveness of the board of directors, the audit committee and auditor quality in

constraining earnings management. The effectiveness of the board and the audit

committee is measured based on size, composition of independent members, financial

expertise and number of meetings. The auditor quality proxies are surrogates by audit

fees, NAS fees and industry specialist auditors. Earnings management is measured by

the absolute value of the discretionary accruals using the Jones model, the modified

Jones model and performance-adjusted discretionary accruals.

The multivariate regression analysis conducted on the sample of 613 firm-year

observations suggests that firms paying higher audit fees and employing the industry

specialist auditors are less likely to manage earnings. These results are robust to

various model specifications, including the 2SLS test. The negative relationship

between audit fees and discretionary accruals may suggest that auditor effort, which

driven by the audit hours, indirectly minimises opportunistic earnings among

managers, due to their concern that such actions may be discovered by the extensive

auditor efforts. This argument is consistent with Caramanis and Lennox (2008), who

suggest that higher audit hours reduce earnings management.

With respect to the auditor independence measures, there is no evidence support that

the NAS fees associates with earnings management. Furthermore, the results for

industry specialist auditors are significant only with respect to the portfolio and

complementary approaches. Previously, Krishnan (2001) has suggested that the

portfolio approach is better suited to capturing the auditors‟ industry expertise,

because some industries which they invest may not be reflected under the market

share approach. The complementary approach, however, captures the complementary

effect s between both the market share and portfolio approaches (Neal and Riley,

2004).

In contrast to the predictions regarding the effectiveness of the board of directors and

audit committee in constraining opportunistic earnings, the present study finds no

evidence that the size, composition, financial expertise or number of meetings affect

the extent of earnings manipulation. It may be due to the monitoring characteristics of

the board and audit committee are offset by the increased auditor quality. The

summary of findings is presented in Table 6.13.

Table 6.13: The summary of the hypothesis and the findings – the relationship

between the corporate governance characteristics‟ and auditor quality in constraining

Hypotheses Findings

committee’s meeting frequency and the engagement

Not supported

size and earnings management. Not supported

H26: There is a negative relationship between the

independent board and earnings management. Not supported

financial expertise and earnings management. Not supported

meeting frequency and earnings management. Not supported

committee’s size and earnings management. Not supported

H30: There is a negative relationship between the solely

independent audit committee and earnings

management.

Not supported

committee’s financial expertise and earnings

management

Not supported

committee’s meeting frequency and earnings

management.

Not supported

H33: There is a negative relationship between audit fees

and earnings management. Supported

H34: There is a positive relationship between NAS fees

and earnings management. Not supported

H35: There is a negative relationship between industry

specialist auditor and earnings management. Supported

CHAPTER 7

SUMMARY AND CONCLUSIONS

7.1 Introduction

This final chapter presents the overview, summary and conclusion of the two

empirical investigations that have been examined in this thesis. The first investigation

looked at the relationship between the characteristics of the board of directors, the

audit committee and audit quality, whilst the second examined the relationship of the

board, the audit committee and external auditor quality in constraining earnings

management. The chapter also details the implications and limitations of the

investigations, as well as suggestions for future research.

7.2 Overview, summary and conclusion of the study

Issues relating to audit quality and earnings management have been the focus of many

scholarly and regulatory debates all around the world. The board of directors, audit

committee and external auditors have been recognised as the mechanisms which,

having the ability to monitor opportunistic earnings and thereby directly linked with

financial reporting quality, consistent with the agency theory proposition.

Unfortunately, previous studies are predominantly US based research where the

litigation environment, governance structure and the auditor reputation are perceived

to be different, thus limit the generalizability of the findings to other countries. This

thesis examines these issues in the context of the UK, based on the FTSE 350 between

the fiscal years 2004 and 2008.

Since investors are unable to directly observe audit quality and earnings management,

they rely on the board of directors, audit committee and auditors to obtain financial

statements that are free from misstatement, error or fraud. Therefore, in this thesis,

there are eight corporate governance characteristics and three proxies of auditor

quality have been empirically examined. Consistent with agency theory and with prior

evidence regarding the effective of certain characteristics of board of directors and

audit committee, the present study posits that the board of director with smaller

number of members, have more independent non-executive directors, possess

financial expertise and have more regular meetings are defined as effective board.

Similarly, an audit committee with more members, comprise solely independent

directors, have more financial expert and meet frequently is also considered as an

effective audit committee. Based on the signalling or reputation hypothesis, audit fees

and industry specialist auditor are used as the proxies of audit quality. In addition, the

NAS is surrogates for auditor independence that have been viewed with scepticism by

the regulators to impair the auditor objectivity while providing the auditing services.

Accordingly, higher audit fees (Abbott et al., 2003a; Carcello et al., 2002; O‟Sullivan,

2000)75

, lower NAS fees (Wines, 1994; Firth, 2002; Frankel et al. 2002;

Raghunandan, 2003; Sharma and Sidhu, 2001; Larcker and Richardson, 2004) and the

engagement of industry specialist auditors (Owhoso et al., 2002; Bédard and Biggs,

1991; O‟Keefe at al., 1994; Carcello and Nagy, 2004) are viewed as a higher audit

quality or higher auditor quality. These characteristics of board and audit committee

and auditor quality proxies are expected to signal to market participants how effective

a given firm is in monitoring financial reporting, therefore conveying the credibility of

the firm‟s financial statements.

Specifically, there are two empirical associations have been examined in this thesis.

First, it examines the association between the effective monitoring characteristics of

board and audit committee on audit quality. In particular, three models of audit quality

are examined: audit fees, NAS fees and industry specialist auditors. From the audit

fees model, the present study finds a positive relationship between audit fees and the

independent non-executive directors on board. This result suggests that independent

board members demand additional and extensive audit effort from auditors in order to

certify their monitoring function, thus increasing audit fees and perceived audit

quality. This result is consistent with the findings of Carcello et al. (2002), Abbot et

al. (2003a), O‟Sullivan (2000) and Adelopo (2010). The other characteristics of the

board and audit committee either marginally correlated or insignificantly correlated

with audit fees. The present study conjectures that these might be due to the

independent board characteristic counteract the other effective characteristics of the

board and audit committee. These results are robust to various model specifications

and tests.

Despite the proposition that higher audit fees are associated with a higher audit quality, consistent

with extensive audit effort and time, the present study is aware of the possibility that lower audit fees

could also be associated with higher audit quality. Refer to Chapter 3 for a detailed argument on this.

The results from NAS fees model yield the opposite proposition, suggesting that a

higher proportion of independent board is associated with higher NAS fees. This

contradicts to the view that independent board use their vigilant oversight function to

limit the NAS as they perceive the higher NAS fees impair the auditor independence.

It is possible that independent board perceive the joint provision of audit and NAS are

not necessarily compromise the auditor independence, but may possibly broaden the

auditors‟ knowledge and improve their audit judgments, resulting in higher audit

quality (see Simunic, 1984; Beck et al., 1988a; Arruñada, 1999a; 1999b; 2000;

Wallman, 1996; Goldman and Barlev, 1974). However, this result is conditional. It is

significant when the levels of NAS (LNNAF) and the sum of the total fees

(LNTOTALFEES) are applied, but no significant evidence is documented when the

NAS ratios are used to measure auditor independence. Prior studies suggest that

LNNAF and LNTOTALFEES are the best measures to capture the economic

importance of the client to the auditor as compared with NAS fees ratios (Ashbaugh et

al., 2003; Larcker and Richadson, 2004). The other corporate governance

characteristics provide inconsistent results to be linked with the NAS fees. These

results are robust to various model specifications and tests.

In association with the auditor industry specialist model, the evidence suggests

inconsistent results between the effectiveness of the board and the audit committee

and their engagement of industry specialist auditors in the year-by-by analysis. In the

pooled sample, four out of five measures of industry specialist auditors suggest that

firms whose audit committee consist solely of independent members and have a lower

number of committee meetings during the year are more likely to employ industry

specialist auditors. Whilst significant, these relationships are, however, sensitive to

the measures of auditor industry expertise and the new variable definition of

independence and meeting frequency of audit committee, thus the present study

caution against drawing inferences from this finding.

The second investigation carried out by this study examines the roles of board of

directors, audit committee and auditor quality in constraining opportunistic earnings.

In order to identify the level of opportunistic earnings behaviour and extreme

reporting decisions made by the management , the absolute value of discretionary

accruals is employed (e.g. Jones 1991; DeFond and Jiambalvo, 1994; Subramanyam

1996; Becker et al., 1998). As in prior studies, the absolute values of discretionary

accruals are estimated using the cross-sectional Jones model and the modified Jones

model as well as performance-adjusted discretionary accruals. As predicted, and

consistent with the prior US studies, the present study finds that firms paying higher

audit fees and engaging industry specialist auditors are likely to be associated with

lower levels of discretionary accruals, suggesting that a higher quality auditor

constrains opportunistic earnings. Firms paying higher audit fees are associated with a

higher auditor effort, thereby minimising the management‟s opportunistic earnings

due to their concern that such actions may be discovered by the extensive auditor

efforts. This proposition is consistent with the prior evidence documented by

Caramanis and Lennox (2008), who suggest a negative relationship between audit

hours (proxy for audit effort) and earnings management. These results are robust to

various model specifications and tests.

In association with auditor independence, there is no evidence to suggest a

relationship between NAS fees and earnings management. This result is contrary to

the evidence reported by Ferguson et al. (2004) and Antle et al. (2006), who suggest

positive and negative relationships between the NAS and earnings management of UK

firms in the periods 1996-1998 and 1994-2000, respectively. This may be due to the

reformation of governance practices in the UK resulting from the revision of the UK

Corporate Governance Code (2010), which first was introduced in July 2003 and

placed major emphasis on the oversight functions of board and audit committees. It

may also be explained by developments in NAS studies. Consistent with this, the first

empirical evidence documented in this thesis suggests that NAS fees are viewed by

independent board members as being able to contribute to a higher quality audit. This

may compensate for the monitoring effects of NAS on opportunistic earnings. As a

result, there is insignificant relationship between NAS and earnings management.

The results of the industry specialist auditor model seem to be sensitive to auditor

industry expertise measures. It suggests a significant relationship with earnings

management when auditor industry specialist is measured using the portfolio and

complementary approaches, but insignificant relationship when the market share

approach is applied. The significant relationship indicates that the industry specialist

auditor has a greater ability to constrain opportunistic earnings than the non-specialist

auditor.

The results for the characteristics of boards and audit committees suggest no evidence

that size, composition, financial expertise and meeting frequency affect the extent of

earnings manipulation. Similarly, the results for ownership structure also suggest an

insignificant relationship with earnings management. These insignificant relationships

may be offset by the monitoring characteristics of a board, audit committee and

institutional investors with the increased in auditor quality. Overall, the results may

suggest that auditors are more effective in constraining opportunistic earnings than

boards of directors, audit committees and institutional investors.

The results on effective characteristics of board and audit committee suggest no

evidence the size of committee, composition of independent members, financial

expertise and number of committee meeting effect the extent of earnings manipulation

in all models. Similarly, the results on the ownership structures also suggest

insignificant relationship with earnings management. These insignificant relationships

may be offset by the effective monitoring characteristics of board, audit committee

and the institutional investor with the increased in auditor quality. Overall, the results

may suggest that the auditors are more effective in constraining the opportunistic

earnings than the board of director, audit committee and institutional investors.

In relation to audit quality measures, both empirical investigations may suggest that

the audit quality surrogates by NAS fees and auditor industry specialist are sensitive

to different type of measures. This may support prior claims that the measurements

for audit quality are complex and problematic (see, for example, Wooten, 2003;

Niemi, 2004; Jensen and Payne, 2005).

Overall, the present study concludes that the findings confirm the proposition of

agency theory on independent board characteristic that certify their monitoring

function by demanding a higher quality audit from the auditors, and that higher

quality auditors have a greater ability to constrain opportunistic earnings than low

quality auditor, resulting in the improvement of financial reporting quality. These

results are generally consistent with the prior studies in this field (see Cohen et al.,

2002; Caramanis and Lennox, 2008).

7.3 Implications of the study

The findings of this thesis should be of potential interest to policy makers,

professionals, the boards of directors and academics, especially on issues relating to

audit quality and corporate governance practice.

Policy makers may use the findings regarding NAS fees to consider the potential

benefits of the joint provision of auditing and NAS. Previously, they claim that NAS

compromises the auditor independence and thus banning the several NAS. In relation

to governance practice, the policy makers should continuously recognise the

important roles played by independent non-executive as one of the fundamental

characteristics of corporate governance system in UK since their monitoring effects

improved the governance systems of the firms.

The professionals, such as financial analysts, may use the findings to integrate the

study on how the market perceived higher audit quality as constraining earnings

management effects the capital market decisions. If the market perceived the firms

with higher audit fees and audited by the industry specialist auditor are associated

with higher financial reporting quality, the reported financial statement may be

viewed as more reliable for investment decision and credit assessment.

The study‟s analysis of corporate governance and audit quality may be of use to

boards of directors as a parameter to estimate how the characteristics of board and the

choice of auditor may influence financial reporting quality. The findings may help

boards of directors to see the positive impact of independent members and higher

quality auditors on audit quality and earnings management.

Finally, the study‟s findings regarding industry specialist auditors may interest

academics, particularly with regard to designing industry specialist auditor measures,

since industry specialist measures are sensitive to the type of approach taken.

7.4 Limitations of the study

This thesis is subject to several potential limitations. Firstly, the sample for this thesis

is drawn from a selection of the larger FTSE 350 UK firms operating in unregulated

industries. Thus, the results of the study may not be applicable to firms below the

FTSE 350 or to regulated firms, because the internal strength of the firms‟ governance

structures varies according to firm size and industry. However, in general, the findings

are in agreement with prior evidence and with agency theory, particularly in relation

to the monitoring function of board of directors and of higher quality auditors.

Secondly, the data used in this thesis are from 2005 to 2008, in which 2008 has been

considered by the economists as the year of global financial crisis due to the collapse

of large financial institutions. There is possibility that the findings may be driven by

the changes in the particular year(s) during or before the financial crisis.

Thirdly, it is possible that some variables may be subjected to some measurement

error. The accruals measures are criticised due to the likelihood of misclassifying the

discretionary and non-discretionary accruals.

Fourthly, the phenomena of earnings management that are indicated in this thesis are

related to the opportunistic earnings. Given that the GAAP practice allowed the

flexibility of accounting choices, the managers may also use their discretion over

earnings to convey the private information, by which could potentially maximise the

firm‟s value. The auditors may underestimate the earnings discretion made by

management since the earnings management involved a higher degree of managerial

judgement. Therefore, the findings in this thesis are restricted on the assumption of

opportunistic earnings rather than the beneficial information.

Fifthly, the audit quality variables may be proxies for something other than is

assumed in the underlying construct of the tests. In this thesis, the audit quality

measures are driven by the auditor‟s reputation capital and perceived auditor

independence, and thus the results are based on market perceptions (audit quality as

perceived by market participants). The use of other audit quality measures such as

restatements and auditors‟ litigation may help to generalise the actual audit quality

rather than the perceived audit quality.

Lastly, there is always the possibility that the models employed in this thesis remain a

potential for certain omitted variables bias that are correlated both with audit quality

and earnings management. However, several steps have been taken to reduce the

likelihood of correlated variables, including the tests for additional control variables,

endogeneity and fixed or random effects models.

Given these limitations, the findings and implications of the study need to be

interpreted with caution.

7.5 Recommendations for future research

There are several ways to extend the studies examined in this thesis. Firstly, as well as

financial expertise characteristic, it is argued that strong industry backgrounds

increase understanding of the business environment, thus helping to improve the

quality of financial reporting. Cohen et al. (2008) argue that audit committee with the

strong industry expertise the will have significant business knowledge and highly

access to the resources that contribute to superior ability to understand and interpret

the business activities and risk. Thus, they able to evaluate whether the firms used

theappropriate reporting procedures, make estimation and assumptions that fit

accordingly to their business environment. Subsequently these may reflect the true

economic value of a given firm, hence enhancing financial reporting quality. Thus,

future research should consider whether financial expertise and strong industry

background make board of directors and the audit committee more effective.

Secondly, as previously noted in the limitations, the results of this thesis are based on

the perceived audit quality measures that driven from the auditors‟ reputation capital

theory. Francis (2004) argues that audit quality can range from low to very high, and

that audit failure can be classified as extremely low audit quality (end quality), which

includes several forms, such as regulatory sanction, litigation, business failure and

earnings restatement. These forms of audit failure are classified as the actual audit

quality measures. Therefore, future studies should investigate how the used of actual

audit quality measures effect the corporate governance and earnings management

could be different compared to the perceived audit quality measures.

Thirdly, the studies on corporate governance and auditor quality in constraining

earnings management can be further examined by taking into account the

complementary and substituting nature of joint effect of both corporate governance

and auditor quality mechanisms. Such research may contribute to the understanding

of the behaviour of auditors and corporate governance mechanisms in association to

the financial reporting quality.

Lastly, the final suggestion for future research can also consider the importance of

corporate voluntary disclosure as a mechanism to limit the opportunistic earnings.

Several studies suggest that the high disclosure quality reduce earnings manipulation

(Jo and Kim, 2007; Lapointe-Antunes et al. 2006). In addition, Beattie (2005) and

Lapointe-Antunes et al. (2006) suggest that the studies of voluntary disclosure and

earnings management are not fully being explored. Thus, the research in this area may

provide comprehensive studies in earnings management.

References

Abbott, L.J., Y. Park, and S. Parker. 2000. The effects of audit committee cctivity and

independence on corporate fraud. Managerial Finance. Vol. 26, No. 11, pp.55-67.

Abbott, L.J., and S. Parker. 2000. Auditor selection and audit committee

characteristics. Auditing: A Journal of Practice & Theory. Vol. 19, No. 2, pp. 47-66.

Abbott, L.J., S. Parker, and G.F. Peters. 2004. Audit committee characteristics and

restatements. Auditing: A Journal of Practice & Theory. Vol. 23, No. 1, pp. 69-89.

Abbott, L.J., S. Parker, G.F. Peters, and K. Raghunandan. 2003a. The association

between audit committee characteristics and audit fees. Auditing: A Journal of

Practice & Theory. Vol. 22, No. 2, pp. 17-32.

Abbott, L.J., S. Parker, G.F. Gary, and K. Raghunandan. 2003b. An empirical

investigation of audit fees, non-audit fees, and audit committees. Contemporary

Accounting Research. Vol. 20, No. 2, pp. 215-234.

Adelopo, I. 2010. The impact of corporate governance on auditor independence: A

study of audit committees in UK listed companies. PhD Thesis, De Montfort

University, Leicester, UK.

Adkins, L. C. and R.C. Hill. 2007. Using Stata for principles of econometrics. Third

edition, John Wiley & Sons, Inc: New York.

Advisory Panel on Auditor Independence. 1994. Public Oversight Board (POB)

Advisory Panel on Auditor Independence, Strengthening the professionalism of the

independent auditor. AICPA: Stamford, Conn.

Anderson, U., K. Kadous, and L. Koonce. 2001. The role of reporting incentives and

quantification in auditors' evaluations of earnings fluctuations, available at

http://ssrn.com/abstract=276570.

Auditing Board Practise (ABP). 2009. International Standard of Auditing, ISA (UK

and Ireland) 200: Overall Objective of the Independent Auditor and the Conduct of an

Audit, available at http://www.frc.org.uk/apb/publications/ pub2076.html.

Auditing Board Practise (ABP). 2004. International Standard of Auditing, ISA (UK

and Ireland) 300: Planning an Audit of Financial Statements, available at

http://www.frc.org.uk/images/uploaded/documents/ISA%20_UK%20and%20Ireland_

%20300.pdf

Aguilera, R.V. C.A. Williams, J.M. Conley and D.E. Rupp. 2006. Corporate

governance and social responsibility: a comparative analysis of the UK and US.

Corporate Governance: An International Review. Vol. 14, No. 3, pp. 147-158.

Agrawal, A., and S. Chadha. 2005. Corporate governance and accounting scandals.

Journal of Law and Economics, Vol. 48, No. 2, pp. 371-406.

AICPA. 2009. Conceptual Framework for AICPA Independence Standards. Vol. 2,

ET sec. 100.01. Code of Professional Conduct. American Institute of Certified Public

Accountant: New York.

Akerlof, G.A. 1970. The market for 'lemons': Quality uncertainty and the market

mechanism, Quarterly Journal of Economics. Vol. 84, No. 3, pp. 488-500.

Allen, F. 1984. Reputation and Product Quality. Rand Journal of Economics. Vol. 15,

No. 3, pp. 311-327.

American Institute of Certified Public Accountants (AICPA). 1978. Commission on

Auditors' Responsibilities: Report, Conclusions and Recommendations. American

Institute of Certified Public Accountant: New York.

Antle, R., E. Gordon, G. Narayanamoorthy, and L. Zhou. 2006. The joint

determinations of audit fees, non-audit fees, and abnormal accruals. Review of

Quantitative Finance and Accounting. Vol. 27, No. 3, pp. 235-266.

Arruñada, B. 1999a. The economics of audit quality: Private incentives and the

regulation of audit and non-audit Services, Kluwer: Dordrecht.

Arruñada, B. 1999b. The provision of non-audit services by auditors: Let the market

evolve and decide. International Review of Law and Economics. Vol. 19, pp. 513-531.

Arruñada, B. 2000. Audit Quality: Attributes, Private safeguards and the role of

Regulation. The European Accounting Review. Vol. 9, No. 2, pp. 205-224.

Arya, A., G. Jonathan, and S. Shyam. 2003. Are unmanaged earnings always better

for shareholders? Accounting Horizons. Vol. 17, pp. 111-116.

Ashbaugh, H., R. LaFond, and B.W. Mayhew. 2003. Do nonaudit services

compromise auditor independence? Further evidence. The Accounting Review. Vol.

78, No. 3, pp. 611-639.

Baber, W., K. Kumar, and T. Verghese. 1995. Client security price reactions to the

Laventhol and Horwath bankruptcy. Journal of Accounting Research. Vol. (Autumn),

pp. 385-395.

Bailey, A. D. And A.A. Grambling. 2005. Financial reporting quality. Research on

Professional Responsibility and Ethics in Accounting. Vol.10, pp. 3-34.

Baker, C.R., and D.M. Owsen. 2002. Increasing the role of auditing in corporate

governance. Critical Perspective on Accounting. Vol. 13, pp. 783-795.

Balsam, S. J. Krishnan, and J.S. Yand. 2003. Auditor industry specialization and

earnings quality. Auditing: A Journal of Practice & Theory. Vol. 22, No. 2, pp. 71-97.

Balvers, R.J., B. McDonald, and R.E. Miller. 1988 Underpricing of new issues and

the choice of auditor as a signal of investment banker reputation. The Accounting

Review. Vol. LXIII, No. 4 (October), pp. 605-622.

Barkess and R. Simnett. 1994. The provision of other services by auditors:

independence and pricing issues. Accounting and Business Research. Vol. 24, pp. 99-

Barth, M.E., J.A. Elliott, and M.W. Finn. 1999. Market rewards associated with

pattern of increasing earnings. Journal of Accounting Research. Vol. 37, No. 2, pp.

387-413.

Bar-Yosef, S. and J. Livnat. 1984. Auditor selection: An incentive-signalling

approach. Accounting and Business Research. Vol. 14, No. 56, pp.301-309

Baum, C.F., M.E. Schaffer, S. Stillman. 2003. Instrumental variables and GMM:

Estimation and Testing. The Stata Journal. Vol. 3, No. 1, pp. 1-31.

Beasley, M.S. 1996. An empirical analysis of the relation between the board of

director composition and financial statement fraud. The Accounting Review. Vol. 71,

No. 4, pp. 443-465.

Beasley, M.S. and K.R. Petroni. 2001. Board independence and audit-firm type.

Auditing: A Journal of Practice & Theory. Vol. 20, No. 1, pp. 97-114.

Beattie, V., and S. Fearnley. 2002. Auditor independence and non-audit services: A

literature review, Institute of Chartered Accountants of England and Wales (ICAEW):

London.

Beaver, W. and E. Engel. 1996. Discretionary behavior with Respect to allowances

for loan losses and behavior of Security Price. Journal of Accounting and Economics.

Vol. 19, pp. 117-206.

Beck, P., T. Frecka, and I. Solomon. 1988a. A model of the market of MAS and audit

services: Knowledge spillovers and auditor-auditee bonding. Journal of Accounting

Literature. Vol. 7, pp. 50-64.

Beck, P.J., T.J. Frecka, and I. Solomon. 1988b. An empirical analysis of the

relationship between MAS involvement and auditor tenure: Implications for auditor

independence. Journal of Accounting Literature. Vol. 7, pp. 65-84.

Becker, C.L., M.L. DeFond, J. Jiambalvo, and K.R. Subramanyam. 1998. The effects

of audit quality on earnings management. Contemporary Accounting Research. Vol.

15, No. 1, pp. 1-24.

Bédard, J.C., and S. Biggs. 1991. The effect of domain-specific experience on

evaluation of management representations in analytical procedures. Auditing: A

Journal of Theory & Practice. Vol. 10, pp. 77-90.

Bédard, J., S.M. Chtourou and L. Courteau. 2004. The effect of audit committee

expertise, independence, and activity on aggressive earnings management. Auditing:

A Journal of Practice & Theory. Vol. 23, No. 2, pp. 13-35.

Berle, A. and G. Means. 1932. The Modern Corporation and Private Property.

Macmillan: New York.

Blue Ribbon Committee (BRC). 1999. Report and recommendations of Blue Ribbon

Committee on improving the effectiveness of corporate audit committees, New York

Stock Exchange and National Association of Securities Dealers: New York.

Boo, E., and D. Sharma. 2008. Effect of regulatory oversight on the association

between internal governance characteristics and audit fees. Journal of Accounting

and Finance. Vol. 48, No. 1, pp. 51-71

Bowen, R.M., E. Noreen, and J. Lacey. 1981. Determinants of the corporate decision

to capitalize interest. Journal of Accounting & Economics. Vol. 3, pp. 151-179.

Bowen, R.M., S. Rajgopal and M. Venkatachalam. 2008. Accounting discretion,

corporate governance, and firm performance. Contemporary Accounting Research.

Vol. 25, No. 2, pp. 351-405.

Box, G. E. P. 1953. Non-normality and tests on variances. Biometrika. Vol. 40, pp.

318-335.

Brennan, N. and M. McDermott. 2004. Alternative perspective on independence of

director. Corporate Governance: An International Review. Vol. 12, No. 3, pp. 325-

Burgstahler, D., and I. Dichev. 1997. Earnings management to avoid earnings

decreases and losses. Journal of Accounting and Economics. Vol. 24, No. 1, pp. 99-

Cadbury Report, The. 1992. Report of the committee on the financial aspects of

corporate governance. Gee Ltd (Professional Publishing Ltd): London.

Cahan, S. R. 1992. The effect of antitrust investigations on discretionary accruals: A

refined test of the political-cost hypothesis. The Accounting Review. Vol. 67, No. 1,

pp. 77-95.

Caminal, R., and X. Vives. 1996. Why market shares matter: An information-based

theory. Rand Journal of Economics. Vol. 27, No. 2, pp. 221-239.

Company Law Reform Bill. 2005. House of Lords, available at http://www.

publications.parliament.uk/pa/ld200506/ldbills/034/2006034.pdf

Caramanis, C. and C. Lennox. 2008. Audit efforts and earnings management. Journal

of Accounting and Economics. Vol. 45, pp. 116-138.

Carcello, J. V., and A.L. Nagy. 2004. Client size, auditor specialization and fraudulent

financial reporting. Managerial Auditing Journal. Vol. 19, No. 5, pp. 651-668.

Carcello, J.V., and T. L. Neal. 2000. Audit committee composition and auditor

reporting. The Accounting Review. Vol. 75, No. 4, pp. 453-467.

Carcello, J.V., D. Hernanson, T. Neal, and R. Riley. 2002. Board characteristics and

audit fees. Contemporary Accounting Research. Vol. 19, No.3, pp. 365-384.

Carcello, J. V. And T.L. Neal. 2003. Audit committee characteristics and auditor

dismissals following “New” going-concern reports. The Accounting Review. Vol. 78,

No. 1, pp. 95-117.

Chaney, P.K, K.L. Philipich. 2002. Shredded reputation: the cost of audit failure.

Journal of Accounting Research. Vol. 40, No. 4, pp.1221-1245.

Chen, G.M. Firth, D.A. Gao, and O.M. Rui. 2006. Ownership structure, Corporate

governance, and fraud: Evidence from China. Journal of Corporate Finance. Vol. 12,

No. 3, pp. 424-448.

Chen, X., P.B. Ender, M. Mitchell and C. Wells. 2003. Regression with Stata,

available at http://www.ats.ucla.edu/stat/stata/webbooks/reg/

Chen, K.Y., and J. Zhou. 2007. Audit committee, board characteristics, and auditor

switch decisions by Andersen's clients. Contemporary Accounting Research. Vol. 24,

No. 4, pp. 1085-1117.

Chen, Y.M., R. Moroney, and K. Houghton. 2005. Audit committee composition and

the use of an industry specialist audit Firm. Accounting and Finance. Vol. 45, No. 2,

pp. 217-239

Chow, C. W. 1982. The demand for external auditing: size, debt and ownership

influences. The Accounting Review. Vol. 57, No. 2, pp. 272-291.

Chtourou, S.M., J. Bédard, and L. Courteau. 2001. Corporate Governance and

Earnings Management, available at http://ssrn.com/abstract=275053.

Chung, R, M. Firth and J. Kim. 2002. Institutional monitoring and opportunistic

earnings management, Journal of Corporate Finance. Vol. 8, pp. 29-48.

Chung, H. And S. Kallapur. 2003. Client importance, nonaudit services, and abnormal

accruals. The Accounting Review. Vol. 78, No. 4, pp. 931-955.

Clarkson, P.M., and D.A. Simunic. 1994. The association between audit quality,

retained ownership, and firm-specific risk in U.S. vs. Canadian IPO markets. Journal

of Accounting and Economics. Vol. 17, pp. 207-228.

Cohen, J., G. Krishnamoorthy, and A. Wright. 2002. Corporate governance and the

audit process. Contemporary Accounting Research. Vol. 19, No. 4, pp. 573-594.

Cohen, J., G. Krishnamoorthy, G. and A. Wright. 2008. Form versus substance: The

implications for auditing practice and research of alternative perspectives on corporate

governance. Auditing: A Journal of Practice & Theory. Vol. 27, No. 2, pp. 181-198

Collier, P., and A. Gregory. 1996. Audit committee effectiveness and the audit fees.

The European Accounting Review. Vol. 5, No. 2, pp. 177-198

Collins, J.H., D.A. Shackelford and J. M.Wahlen. 1995. Bank differences in the

coordination of regulatory capital, earnings, and taxes. Journal of Accounting

Research. Vol. 33, pp. 263-291.

Conyon, M.J., and K.J. Murphy. 2000. The price and the pauper? CEO pay in the

United States and United Kingdom. The Economic Journal. Vol. 110, No. 467, pp.

640-671.

Conger, J., D. Finegold, and E. Lawler III. 1998. Appraising boardroom performance.

Harvard Business Review. Vol. 76, pp. 136-148.

Copley, P.A. and E.B. Douthett. 2002. The association between auditor choice,

ownership retained, and earnings disclosure by firms making initial public offerings.

Contemporary Accounting Research. Vol. 19, No. 1, pp. 49-75.

Craswell, A.T. 1999. Does the provision of non-audit services impair auditor

independence?. Journal of International Auditing. Vol. 3, pp. 29-40.

Craswell, A.T., J.R. Francis and S.L. Taylor. 1995. Auditor brand name reputations

and industry specializations. Journal of Accounting and Economics. Vol. 20, pp. 297-

Craswell, A.T and J.R. Francis. 1999. Pricing of initial audit engagements: a test of

competing theories. The Accounting Review. Vol. 74, pp. 201-216.

Cronqvist, H. and R. Fahlenbrach. 2009. Large shareholders and corporate policies,

Review of Financial Studies. Vol. 22, No. 10. pp. 3941-3976.

Datar, S.M., G.A. Feltham, and J.S. Hughes. 1991. The role of audits and audit

quality in valuing new issues. Journal of Accounting and Economics. Vol. 14, pp. 3-

Davidson, R., J. Goodwin-Stewart, and P. Kent. 2005. Internal governance structures

and earnings management. Accounting and Finance. Vol. 45, No. 2, pp. 241-267.

Davis, J.H., F.D. Schoorman, and L. Donalson. 1997. Toward a stewardship theory of

management. Academy of Management Review. Vo. 222, No. 1, pp. 20-47.

DeAngelo, L. 1981. Auditor size and audit quality. Journal of Accounting and

Economics. Vol. 3, No. 3, pp. 183-199.

DeAngelo, L., 1986. Accounting numbers as market valuation substitutes: a study of

management buyouts of public shareholders. The Accounting Review. Vol. 61, pp.

400–420.

DeAngelo, H., L. DeAngelo, and D. J. Skinner. 1994. Accounting choice in troubled

companies. Journal of Accounting and Economics. Vol. 17, pp. 113-143

Dechow, P.M, R.G. Sloan, and A.P. Sweeney. 1995. Detecting earnings management.

The Accounting Review. Vol. 70, No. 2, pp. 193-225.

Dechow, P.M, R.G. Sloan, and A.P. Sweeney. 1996. Causes and consequences of

earnings manipulation: An analysis of firms subject to enforcement actions by the

SEC. Contemporary Accounting Research, Vol.13, No.1, pp. 1-36.

DeFond, M.L. 1992. The association between changes in client firm agency costs and

auditor switching. Auditing: A Journal of Practice & Theory. Vol. 11, No. 1, pp. 49-

DeFond, M.L. and J.R. Francis. 2005. Audit research after Sarbanes-Oxley. Auditing:

A Journal of Practice & Theory. Vol. 24, pp. 5-30.

DeFond, M.L and J. Jiambalvo. 1994. Debt covenant violation and manipulation of

accruals. Journal of Accounting and Economics. Vol. 17, No. 1-2, pp. 145-176.

DeFond, M.L. and C.W. Park. 1997. Smoothing income in anticipation of future

earnings. Journal of Accounting and Economics. Vol. 23, No. 2, pp. 115-139.

DeFond, M.L., J.R. Francis and T.J. Wong. 2000. Auditor industry specialization and

market segmentation: Evidence from Hong Kong. Auditing: A Journal of Practice &

Theory. Vol. 19, No. 1, pp. 49-66.

DeFond, M., K. Raghunandan, and K. R. Subramanyam. 2002. Do non-audit services

impair auditor independence? Evidence from going-concern audit opinions. Journal

of Accounting Research. Vol. 40, No. 4, pp. 1247-1274.

DeFond, M.L., R.N. Hann and X. Hu. 2005. Does the market value financial expertise

on audit committees of boards of directors? Journal of Accounting Research. Vol. 43,

No. 2, pp. 153-193.

Degeorge, F., J. Patel, and R. Zeckhauser. 1999. Earnings management to exceed

thresholds. Journal of Business. Vol. 72, pp. 1-33.

Deis, D.R., and G. Giroux. 1996. The effect of auditor changes on audit fees, audit

hours, and audit quality. Journal of Accounting and Public Policy. Vol. 15, No. 1, pp.

55-76.

Denis, D.K. Twenty-five years of corporate governance research... and counting.

Review of Financial Economics. Vol. 10, pp. 191-212.

DeZoort, F.T. 1998. An investigation of audit committees‟ oversight responsibilities.

ABACUS. September, pp. 208-227.

DeZoort, F.T., D.R. Hermanson, D.S. Archambeault, and S.A. Reed. 2002. Audit

committee effectiveness: a synthesis of the empirical audit committee literature.

Journal of Accounting Literature. Vol. 21, pp. 38-75.

Dhaliwal, D., G. Salamon, and E. Smith. 1982. The effect of owner versus

management control on the choice of accounting methods. Journal of Accounting &

Economics. Vol. 4, pp. 41-53.

Donaldson, L., and J. H. Davis. 1991. Stewardship theory or agency theory: CEO

governance and shareholder returns. Australian Journal of Management. Vol. 16, pp.

49-64.

Donaldson, T., and L.E. Preston. 1995. The stakeholder theory of the corporation:

concepts, evidence, and implications. Academy of Management Review. Vol. 20, No.

1, pp. 65-91.

Dopuch, N. and D. Simunic. 1982. Competition in auditing: an assessment.

Symposium on Auditing Research Vol. IV, pp. 401-450

Duff, A. 2004. AUDITQUAL: Dimensions of Audit Quality, Institute of Chartered

Accountants of Scotland: Edinburgh.

Dunn, K.A. and B.W. Mayhew. 2004. Auditor industry specialization and client

disclosure quality, Review of Accounting Studies. Vol. 9, No.1, pp. 35–58.

Dunn, K.A., B.W. Mayhew, and S. Morsefield. 2000. Disclosure quality and auditor

choice. Working paper CUNY-Baruch.

Dye, R. A.1993. Auditing standards, legal liability, and auditor wealth. The Journal of

Political Economy. Vo. 101, No. 5, pp. 887-914.

Eisenberg, T., S. Sudgren and M. Wells. 1998. Larger board size and decreasing firm

value in small firms. Journal of Financial Economics, Vol. 48, No. 1, pp. 35-54.

Elitzur R. and H. Falk. 1996. Planned audit quality. Journal of Accounting and Public

Policy. Vol. 15, No. 1, pp. 247-269.

Fama, E.F., and M. C. Jensen. 1983. Separation of ownership and control. Journal of

Law and Economics. Vol. 26, pp. 301-325.

Ferguson, A., and D. Stokes. 2002. Brand name audit pricing, industry specialisation

and leadership premiums post Big 8 and Big 6 mergers. Contemporary Accounting

Ferguson, M.J., G.S. Seow, and D. Young. 2004. Nonaudit Services and Earnings

Management: UK Evidence. Contemporary Accounting Research, Vol. 21, No. 4, pp.

813-841.

Fich, E. and A. Shivdasani. 2007. Financial fraud, director reputation, and shareholder

wealth. Journal of Financial Economics. Vol. 86, pp. 306-336.

Fields, T., T. Lys, and L. Vincent. 2001. Empirical research on accounting choice.

Journal of Accounting and Economics. Vol. 31, pp. 255-307.

Firth, M. 1980. Perceptions of auditor independence and official ethical guidelines.

The Accounting Review. Vol. LV, No.3, pp.451-466.

Firth, M. 1997. The provision of non-audit services by accounting firms to their

clients. Contemporary Accounting Research. Vol. 14, No. 2, pp. 1-21.

Firth, M. 2002. Auditor-provided consultancy services and their associations with

audit fees and audit opinions. Journal of Business and Accounting. Vol. 29, No. 5 and

6, pp. 661-693.

Firth, M. and A. Smith. 1992. Selection of auditor firms by companies in the new

issue market. Applied Economics. Vol. 24, pp. 247-255.

Francis, J.R., E.L. Maydew, and H.C. Sparks. 1999. The role of big 6 auditors in the

credible reporting of accruals. Auditing: A Journal of Practice & Theory. Vol. 18, No.

2, pp. 17-34.

Francis, J.R., and E.R. Wilson. 1988. Auditor changes: A join test of theories relating

to agency costs and auditor differentiation. The Accounting Review. Vol. 63, No. 4,

pp. 663-682.

Francis, J. and D. Simon. 1987. A test of audit pricing in the small-client segment of

the U.S. audit market. The Accounting Review. Vol. 62, pp.145-157.

Francis, J. and D. Stokes. 1986. Audit prices, product differentiation, and scale

economies: Further evidence from the Australian audit market. Journal of Accounting

Francis, J.R. 2004. What do we know about audit quality? The British Accounting

Review. Vol. 36, pp.15-36.

Francis J.R. 2006. Are auditors compromised by nonaudit services? Assessing the

evidence. Contemporary Accounting Research. Vol. 23, No. 3, pp. 747-760.

Frankel, R.M., M.F. Johnson, and K.K. Nelson. 2002. The relation between auditors‟

fees for nonaudit services and earnings management. The Accounting Review. Vol. 77,

pp. 71-105.

Financial Reporting Council (FRC). 2008. The audit quality framework. Financial

Reporting Council: London.

Freeman, R.E. 1984. Strategic management: A stakeholder approach. Pitman: Boston.

Gaver, J.J., K.M. Gaver and J. R. Austin. 1995. Additional evidence on bonus plans

and income management. Journal of Accounting and Economics. Vol. 19, pp. 3-28.

Gilson, S.C. 1990. Bankruptcy, boards, banks, and blockholders: Evidence on changes

in corporate ownership and control when firms default. Journal of Financial

Glass, G.V. and K.D. Hopkins. 1996. Statistical methods in education and

psychology (3rd ed.). Allyn and Bacon: Boston, MA.

Glass, G.V, P.D. Peckham, and J.R. Sanders. 1972. Consequences of failure to meet

the assumptions underlying the fixed effects analysis of variance and covariance.

Review of Educational Research. Vol. 42, pp. 237-288.

Goddard A. and C. Masters. 2000. Audit committees, Cadbury code and audit fees: an

empirical analysis of UK companies. Managerial Auditing Journal. Vol. 15, No. 7,

pp. 358-371.

Goldman, A. and B. Barlev. 1974. The Auditor-Firm Conflict of Interests: Its

Implications for Independence. The Accounting Review, October, pp. 707-718.

Griffin, P.A. D.H. Lont, and Y. Sun. 2008. Corporate governance and audit fees:

Evidence of countervailing relations. Journal of Contemporary Accounting &

Gujarati, D. 2003. Basic econometrics (5th ed.). McGraw-Hill/Irwin: New York.

Habbash, M. 2010. The effectiveness of corporate governance and external audit on

constraining earnings management practice in the UK. PhD Thesis, Durham

University, UK.

Habbash, M., A. Salama, R. Dixon, and K. Hussainey. 2010. The effects of non-

executive directors‟ commitment, chairman independent and ownership structure on

earnings management. Journal of Applied Accounting Research, Vol. 13ii,

forthcoming.

Habib, A., and A. Islam. 2007. Determinants and consequences of non-audit service

fees: Preliminary evidence from Bangladesh. Managerial Auditing Journal. Vol. 22,

No. 5, pp. 446-469.

Hair, J.F., W.C. Black, B.J. Babin, and R.E. Anderson. 2010. Multivariate data

analysis: A global perspective (7th ed.), Pearson Prentice Hall: Upper Saddle River,

New Jersey.

Hamilton, L.C. 1992. Regression with graphics. Wadsworth: Belmont, CA.

Han, J.C.Y., and S-W. Wang. 1998. Political Costs and Earnings Management of Oil

Companies during the 1990 Persian Gulf Crisis. The Accounting Review. Vol. 73, No.

1, pp. 103-117.

Hartley, R.V. and T.L. Ross. 1972. MAS and audit Independence: An image problem.

Journal of Accountancy. November, pp. 42-51.

Healy, P.M. and J.M. Wahlen.1999. A review of the earnings management literature

and its implications for standard setting. Accounting Horizons. Vol. 13, No. 4, pp.

365-383.

Healy, P.M. 1985. The impact of bonus schemes on the selection of accounting

principles. Journal of Accounting and Economics. Vol. 7, pp. 85-107.

Heninger, W.G. 2001. The association between auditor litigation and abnormal

accruals. The Accounting Review. Vol. 76, No. 2, pp. 111-126.

Higgs Report. 2003. Review of the role and effectiveness of non-executive directors.

Department of Trade and Industry: London.

Hillison, W. and C. Pacini. 2004. Auditor reputation and the insurance hypothesis:

The information content of disclosures of financial distress of a major accounting

firm. Journal of Managerial Issues. Vol. 16, No. 1, pp. 65-86.

Hillson, W. and M. Kennelley, M. 1988. The economics of nonaudit services.

Accounting Horizons. Vol. 2, No. 3, pp. 32-40.

Hribar, P. and D.W. Collins. 2002. Error in Estimating Accruals: Implication for

Empirical Research. Journal of Accounting Research. Vol. 40, No. 1, pp. 105-134.

Humphrey, C. 1997, Debating audit expectation. Current Issues in Auditing (M.

Sherer and S. Turley, editors). Paul Chapman Publishing Ltd. London.

Hogan, C. 1997. Costs and benefits of audit quality in the IPO market: A self-

selection analysis. The Accounting Review. January, pp. 67-86.

Hoitash, R., A. Markelevich, and C.A. Barragato. 2007. Auditor fees and audit

quality. Managerial Auditing Journal. Vol. 22 No. 8, pp. 761-786.

Holthausen, R.W., D.F. Larkers and R.G. Sloan. 1995. Annual bonus schemes and the

manipulation of earnings. Journal of Accounting and Economics. Vol. 19, pp. 29-74.

Houghton, K. A. and C.C. Ikin. 2001. Auditor provided non-audit services: modelling

fees and willingness to buy. Working paper, University of Tasmania, Hobart.

Hung, H. 1998. A typology of the theories of the roles of governing boards.

Corporate Governance: An International Review. Vol. 6, No. 2, pp. 101-111.

Institute of Chartered Accountants in England and Wales (ICAEW). 2002. Audit

Quality (Abridged). ICAEW: London.

Imhoff, E.A. 2003. Accounting quality, auditing and corporate governance.

Accounting Horizons. Vol. 13, pp. 117-128.

Jensen, M.C. 1993. The modern industrial revolution, exit, and the failure of internal

control systems. The Journal of Finance. Vol. 48, No. 3, pp. 831-880.

Jensen, M.C. and W.H. Meckling. 1976. Theory of the firm: managerial behavior,

agency costs and ownership structure. Journal of Financial Economics. Vol. 3, No. 4,

pp. 305-360.

Jensen, K., and J. Payne. 2005. Audit procurement: managing audit quality and audit

fees in response to agency costs. Auditing: A Journal of Practice & Theory. Vol. 24,

No. 2, pp. 27-48.

Johnson, P., K. Jamal, and R. Berrman. 1991. Effects of framing on auditor decisions.

Organization Behaviour and Human Decision Process, Vol. 50, pp. 72-105.

Johnson, W.B. and T. Lys. 1990. The market for audit services: Evidence from

voluntary auditor changes. Journal of Accounting and Economics. Vol. 12, No. 1-3,

pp. 281-308.

Jones, J. 1991. Earnings management during import relief investigations. Journal of

Kang, S. And K. Sivaramakrishnan. 1995. Issues in testing earnings management and

an instrumental variable approach. Journal of accounting Research. Vol. 33, No. 2,

pp. 353-367.

Khurana, I.K. and K.K. Raman. 2004. Litigation risk and the financial reporting

credibility of big 4 versus non-big 4 audits: Evidence from Anglo-American

countries. The Accounting Review. Vol. 79, No. 2, pp. 473-495.

Khurana, I.K. and K.K. Raman. 2006. Do investors care about the auditor‟s economic

dependence on the client? Contemporary Accounting Research. Vol. 23, No. 4, pp.

977-1016.

Kinney, W.R., Z-V. Palmrose and S. Scholz. 2004. Auditor independence, non-audit

services, and restatements: Was the U.S. government right? Journal of Accounting

Research. Vol. 42, No. 3, pp. 561-588.

Kim, J.B., R. Chung, and M. Firth. 2003. Auditor conservatism, asymmetric

monitoring, and earnings management. Contemporary Accounting Research. Vol. 20,

No. 2, pp. 323-359.

Kirchmaier, T., G. Owen, and. J. Grant. 2005. Corporate governance in the US and

Europe: Where are we now?, available at http://ssrn.com/abstract=708461.

Klein, B., and K.B. Leffler. 1981. The role of market forces in assuring contractual

performance. Journal of Political Economy. Vol. 98, pp. 615-641.

Klein, P. 2002. Audit committee, board of director characteristics, and earnings

management. Journal of Accounting and Economics. Vol. 33, No. 3, pp. 375-400.

Knapp, M.C. 1991. Factors that audit committee members use as surrogates for audit

quality. Auditing: A Journal of Practice & Theory. Vol. 10, No. 1, pp. 35-52.

Knechel, W.R, V. Naiker, and G. Pacheco. 2007. Does auditor industry specialization

matter? Evidence from market reaction to auditor switches. Auditing: A Journal of

Practice & Theory. Vol. 26, No. 1, pp. 19-45.

Kothari, S.P., A.J. Leone, and C.E. Wasley. 2005. Performance matched discretionary

accruals. Journal of Accounting and Economics. Vol. 39, pp. 163-197.

Krishnan, G.V. 2003a. Does big 6 auditor industry expertise constrain earnings

management. Accounting Horizons. Supplement, pp. 1-16.

Krishnan, G.V. 2003b. Audit quality and the pricing of discretionary accruals.

Auditing: A Journal of Practice & Theory. Vol. 22, No. 1, pp. 109-126.

Krishnan, J. A. 2001. A comparison of auditors‟ self-reported industry expertise and

alternative measures of industry specialisation. Journal of Accounting and

Krishnan, J. and J.E. Lee. 2009. Audit committee expertise, litigation risk and

corporate governance. Auditing: A Journal of Practice & Theory. Vol. 28, No. 1, pp.

241-261.

Krishnan, G.V., and G. Visvanathan. 2008. Does the SOX definition of an accounting

expert matter? The association between audit committee accounting expertise and

accounting conservatism. Contemporary Accounting Research. Vol. 25, No. 3,

pp.827-857.

Krishnan, G.V., and G. Visvanathan. 2009. Do auditors price audit committee‟s

expertise? The case of accounting versus nonaccounting financial experts. Journal of

accounting, Auditing and Finance. Vol. 24, No. 1, pp. 115-144.

Kwon, S.Y., C.Y. Lim and P.M. Tan. 2007. Legal systems and earnings quality: The

role of auditor industry specialization. Auditing: A Journal of Practice & Theory. Vol.

26, No. 2, pp. 25-55.

Lai, K-W. and G.V. Krishnan. 2009. Are non-audit services associates with firm

value? Evidence from financial information system-related services. Accounting and

Finance. Vol. 49, pp. 599-617.

Lai, K-W., and F.A. Gul. 2008. Was audit quality of Laventhol and Horwath poor?

Journal of Accounting and Public Policy. Vol. 27, No. 3, pp. 217-237.

Larcker, D. F. and S.A. Richardson. 2004. Fees paid to audit firms, accrual choices,

and corporate governance. Journal of Accounting Research. Vol. 42, No. 3, pp. 625-

Larcker, D.F. and T.O. Rusticus. 2010. On the use of instrumental variables in

accounting research. Journal of Accounting and Economics. Vol. 49, No. 3, pp. 186-

Lin, J.W and M. I. Hwang. 2009. Audit quality, corporate governance and earnings

management: a meta analysis. International Journal of Auditing, Vol. 14, No. 1, pp.

57-77.

Lavin, D. 1976. Perceptions of the independence of the auditor. The Accounting

Review. Vol. 51, No. 1, pp. 41-50.

Lavin, D. 1977. Some effects of the perceived independence of the auditor.

Accounting, Organizations and Society. Vol. 2, No. 3, pp. 237-244.

Lawrence, J.J., and G.P. Stapledon. 1999. Is board composition important? A study of

listed Australian companies, available at http://ssrn.com/abstract=193528

Lee, H.Y. 2008. The association between audit committee and board of director

effectiveness and changes in the nonaudit fee ratio. Applied Financial Economics.

Vol. 18, No. 8, pp.629-638.

Lee, H.Y., and V. Mande. 2005. The relationship of audit committee characteristics

with endogenously determined audit and non-audit fees. Quarterly Journal of

Business and Economics. Vol. 44, No. 3/4, pp. 93-112.

Lennox, C.S. 1999. Audit quality and auditor size: An evaluation of reputation and

deep pockets hypotheses. Journal of Business Finance & Accounting. Vol. 26, No. 7-

8, pp. 779-805.

Lennox, C.S. 2003. Opinion Shopping and the Role of Audit Committees when Audit

Firms are Dismissed: The US Experience. The Institute of Chartered Accountants of

Scotland: Edinburgh.

Leuz, C., D. Nanda and P.D. Wysocki. 2003. Earnings Management and Investor

Protection: An International Comparison. Journal of Financial Economics. Vol. 69,

pp. 505-527.

Levitt, A. 1998. The numbers game. Remarks delivered at the NYU Centre for Law

and Business, NY, September 28, available at: http://www.sec.gov/

news/speech/speecharchive/1998/spch220.txt.

Levitt, A. 2000. Renewing the covenant with investors. Remarks delivered at the

NYU Center for Law and Business, NY, May 10, available at: http://www.sec.

gov/news/speech/spch370.htm.

Lin, Z.J. M. Liu., and Z. Wang. 2009. Market implications of the audit quality and

auditor Switches: Evidence from China. Journal of International Financial

Management and Accounting. Vo. 20, No. 1, pp. 35-78

Linberg, D. L. 2001. Discussion of the demand for auditor reputation across

international markets for audit services. The International Journal

of Accounting. Vol. 36, No. 4, pp. 429-432

Lipton, M. And J.W. Lorsch. 1992. A modest proposal for improved corporate

governance. The Business Lawyer. Vol. 48, pp. 59-77.

Louis, H. and D. Robinson. 2005. Do managers credibly use accruals to signal private

information? Evidence from the pricing of discretionary accruals around stock splits.

Low, K. Y. 2004. The effects of industry specialization on audit risk assessments and

audit-planning decisions. The Accounting Review. Vol. 79, pp. 201-219.

Lys, T., and R.L. Watts. 1994. Lawsuits against auditors. Journal of Accounting

Matsumoto, D. 2002. Management's incentives to avoid negative earnings surprises.

The Accounting Review. Vol. 77, No. 3, pp. 483-514.

McMullen, D.A. 1996. Audit Committee Performance: An investigation of the

consequences associated with audit committees. Auditing: A Journal of Practice &

Theory. Vol. 15, No. 1, pp. 87-103.

McNichols, M.F. and G.P. Wilson. 1988. Evidence of earnings management from the

provision for bad debts. Journal of Accounting Research. Vol. 26, pp. 1-31.

McNichols, M.F. 2000. Research Design Issues in Earnings Management Studies.

Journal of Accounting and Public Policy. Vol. 19, pp. 313-345.

Menard, S. 2002. Applied logistic regression analysis, (2nd Ed.), Sage University

Papers Series on Quantitative Applications in the Social Sciences, Series No. 07-106,

Sage: Newbury Park, CA.

Menon, K., and J. Williams. 1994. The use of audit committees for monitoring.

Journal of Accounting and Public Policy. Vol. 13, No. 2, pp. 121-139.

Menon, K. and D.D. Williams. 2001. Long-term trends in audit fees. Auditing: A

Journal of Practice & Theory. Vol. 20, No. 1, pp. 115-136.

Moses, O. D. 1987. Income smoothing and incentives: Empirical tests using

accounting changes. The Accounting Review. Vol. LXII, No. 2, pp. 358-377.

Moizer, P. 1992. State of the art in audit market research. European Accounting

Review. Vol. 1, pp. 333-348.

Moizer, P. 1997. Auditor reputation: The international empirical evidence.

International Journal of Auditing. Vol. 1, No. 1, pp. 61-74.

Monks, R. A. G., and N. Minow. 2008. Corporate governance. John Wiley: West

Sussex, UK.

Muth, M.M., and L. Donaldson. 1998. Stewardship theory and board structure: A

contingency approach. Corporate Governance: An International Review. Vol. 6, No.

1, pp. 5-28.

Neal, T.L and R.R. Riley. 2004. Auditor industry specialist research design. Auditing:

A Journal of Practice & Theory. Vol. 23, No. 2, pp. 169-177.

Nichols, D.R. and K.H. Price. 1976. The auditor-firm conflict: An analysis using

concepts of exchange theory. The Accounting Review. Vol. 51, pp. 335-346.

Niemi, L. 2004. Auditor size and audit pricing: evidence from small audit firms. The

European Accounting Review. Vol. 13, No. 3, pp. 541-560.

O‟Keefe, T.B., D.A. Simunic and M.T. Stein. 1994. The production of audit services:

Evidence from a major public accounting firm. Journal of Accounting Research. Vol.

32, No. 2, pp. 241-261.

O‟Reilly, D., R. Leitch, and B. Tuttle. 2006. An experimental test of the interaction of

the insurance and information-signaling hypotheses in auditing. Contemporary

Osma, B.G. and B.G. Noguer. 2007. The effect of the board composition and its

monitoring committees on earnings management: Evidence from Spain. Corporate

Governance: An International Review. Vol. 15, No. 6, pp. 1413-1428.

O‟Sullivan, N. 2000. The impact of board composition and ownership on audit

quality: Evidence from large UK companies. The British Accounting Review. Vol. 32,

No. 4, pp. 397-414.

O‟Sullivan, N. 1999. Board Characteristics and audit pricing post-Cadbury: A

research note. European Accounting Review. Vol. 8, No. 2, pp. 253-263.

O‟Sullivan, N., and S.R. Diacon. 2002. The impact of ownership, governance and

non-audit services on audit fees: Evidence from the insurance industry. International

Journal of Auditing. Vol. 6, No. 1, pp. 93-107.

Owhoso, V. E., W. F. Messier and J. G. Lynch. 2002. Error detection by industry

specialized teams during sequential audit review. Journal of Accounting

Palmrose, Z-V. 1988. An analysis of auditor litigation and audit service quality. The

Accounting Review. Vol. 63, pp. 55-73.

Palmrose, Z-V. 1986a. Audit fees and auditor size: Further evidence. Journal of

Palmrose, Z-V. 1986b. The effect of non-audit services on the pricing of audit

services: Further evidence. Journal of Accounting Research. Vol. 24, No. 2, pp. 405-

Palmrose, Z-V. 1991. Trials of legal disputes involving independent auditors: Some

empirical evidence. Journal of Accounting Research. Vol. 29, No. 3, pp. 149-185.

Palmrose, Z-V. and S. Scholz. 2004. The circumstances and legal consequences of

Non-GAAP reporting: Evidence from restatement. Contemporary Accounting

Pampel, F.C. 2000. Logistic Regression: A Primer, Sage University Papers Series on

Quantitative Applications in the Social Sciences, Series No. 07-132, Sage: Newbury

Park, CA.

Panel on Audit Effectiveness. 2000. The panel on audit effectiveness report and

recommendations. Public Oversight Board: Stamford, CT.

Pany, K. And P.M.J. Reckers.1988. Auditor performance of MAS: A study of its

effects on decisions and perceptions. Accounting Horizons. Vol. 2, pp. 31-38.

Park, Y.W., and H.H. Shin. 2004. Board Composition and earning Management in

Canada. Journal of Corporate Finance. Vol. 10, No. 3, pp. 431-457

Parkash, M. and C.F. Venable. 1993. Auditee incentives for auditor independence:

The case of nonaudit services. The Accounting Review. Vol. 68, No. 1, pp. 113-133.

Public Company Accounting Oversight Board (PCAOB). 2008. Discussion- Treasury

Advisory Committee‟s Recommendation Relating to the feasibility of developing key

indicators of audit quality and effectiveness, available at http://pcaobus.org/

News/Events/Documents/10222008_SAGMeeting/BP_Feasability _AQI.pdf

Peasnell,K.V., P.F. Pope and S. Young. 2000. Accrual management to meet earnings

targets: UK evidence pre- and post-Cadbury. British Accounting Review. Vol. 32, pp.

415-445.

Peasnell, K.V., P.F. Pope and S. Young. 2005. Board monitoring and earnings

management: Do outside directors influence abnormal accruals?. Journal of Business

Finance & Accounting. Vol. 32, No. 7/8, pp. 1311-1346.

Philips, J., M. Pincus and A.O. Rego. 2003. Earnings management: New evidence

based on deferred tax expense. The Accounting Review. Vol. 78, No. 2, pp. 491-521.

Power, M. 1999. The audit society: rituals of verification. Oxford University Press:

New York.

Pratt, J. and J. D. Stice. 1994. The effects of client characteristics on auditor litigation

risk judgments, required audit evidence, and recommended audit fees. The Accounting

Review. Vol. 69, No. 4, pp. 639-656.

Press, E., and J. Weintrop.1 990. Accounting-based constraints in public and private

debt agreements: Their association with leverage and impact on accounting choice.

Raghunandan, K. 2003. Non-audit services and shareholder ratification of auditors

appointments. Auditing: A Journal of Practice & Theory. Vol. 22, No. 1, pp. 155-

Rogerson, W.P. 1983. Reputation and product quality. The Bell Journal of

Ronen, J. and V. Yaari. 2008. Earnings management emerging insights in theory,

practice and research. Springer: New York.

Ruddock, C., S.J. Taylor and S.L. Taylor. 2006. Nonaudit Services and Earnings

Conservatism: Is Auditor Independence Impaired?. Contemporary Accounting

Ryan, S.G., R.H. Herz, T.E. Iannaconi, L.A. Maines, K. Palepu. C. M. Schrand, D.J.

Skinner and L. Vincent. 2001. SEC auditor independence requirements. Accounting

Horizons. Vol. 15, No. 4, pp. 373-386.

Sankar, M.R., and K.R. Subramanyam. 2001. Reporting discretion and private

information communication through earnings. Journal of Accounting Research. Vol.

99, No. 2, pp. 365-386.

Sarbanes-Oxley Act (SOX). 2002. Public Law No. 107-204. GPO: Washington, DC.

Schipper, K. 1989. Commentary on earnings management. Accounting Horizons, Vol.

3, No. 4, pp. 91-102.

Schwartz, K., and K. Menon. 1985. Auditor switches by failing firms. The Accounting

Review. Vol. 60, No.2, pp.248-61.

Securities and Exchange Commission (SEC). 2000. Final Rule: Revision of the

Commission‟s Auditor Independence Requirements, FRR No. 56. Washington, DC.

Shapiro, C. 1983. Premiums for high quality products as returns to reputation.

Quarterly Journal of Economics. Vol. 98, pp. 659-679.

Sharma, D., and J. Sidhu. 2001. Professionalism vs commercialism: The association

between non-audit services (NAS) and audit independence. Journal of Business

Finance and Accounting. Vol. 28, No. 5-6, pp. 563-594.

Sharma, V.D., V. Naiker, and B. Lee. 2009. Determinants of audit committee meeting

frequency: evidence from a voluntary governance system. Accounting Horizons. Vol.

23, No. 3, pp. 245-263.

Shakun, M.F. 1978. Cost-Benefit Analysis of Auditing. Research Study No. 3,

Commission on Auditors‟ Responsibilities. American Institute of Certified Public

Accountants, Inc.: New York.

Shleifer, A., and R. Vishny. 1997. Large shareholders and corporate control. Journal

of Political Economy. Vol. 94, pp. 96 1-88.

Shockley, R.A. and R.N. Holt. 1983. A behavioral investigation of supplier

differentiation in the market for audit services. Journal of Accounting Research. Vol.

21, No. 2, pp. 545-564.

Simunic, D.A. 1980. The pricing of audit services: theory and evidence. Journal of

Simunic, D.A. 1984. Auditing, consulting, and auditor independence. Journal of

Accounting Research. Vol. 22, No. 2, pp. 679-702

Simunic, D. A. and M.T. Stein. 1987. Product differentiation in auditing: Auditor

choice in the market for unseasoned new issues. Monograph Prepared for the

Canadian Certified General Accountant Research Foundation.

Simunic, D.A., and M.T. Stein. 1996. The impact of litigation risk on audit pricing: A

review of the economics and the evidence. Auditing: A Journal of Practice & Theory.

Vol. 15, pp. 119-134.

Skinner, D. J. and R.G. Sloan. 2002. Earnings surprises, growth expectations and

stock returns or don‟t let an earnings torpedo sink your portfolio. Review of

Accounting Studies. Vol. 7, pp. 289-312.

Smallwood, D., and J. Conlisk. 1979. Product quality in markets where consumers are

imperfectly informed. Quarterly Journal of Economics. Vol. 93, No.1, pp. 1-23.

Solomon, J.F. 2007. Corporate Governance and Accountability, (2nd ed.), Wiley:

New York.

Stice, J. 1991. Using financial and market information to identify pre-engagement

factors associated with lawsuits against auditors. The Accounting Review. Vol. 66, pp.

516-533.

St. Pierre, K. and J. Anderson. An analysis of the factors associated with lawsuits

against public accountants. The Accounting Review. Vol. 54, pp. 242-263.

Subramanyam, K.R. 1996. The pricing of discretionary accruals. Journal of

Accounting and Economics. Vol. 22, pp. 249-281.

Sun, N., A. Salama, K. Hussainey and Habsah, M. 2010. Corporate environmental

disclosure, corporate governance and earnings management. Managerial Auditing

Journal, Vol. 25, No. 7, pp. 679-700.

Sundaramurthy, C., and M. Lewis. 2003. Control and collaboration: Paradoxes of

governance. Academy of Management Review. Vol. 28, No. 3, pp. 397-415.

Sweeney, A. 1994. Debt covenant violations and managers‟ accounting responses.

Journal of Accounting and Economics. Vol. 17, No. 3, pp. 281-308.

Taylor, M. H. 2000. The effects of industry specialization on auditors' inherent risk

assessments and confidence judgments. Contemporary Accounting Research. Vol. 17,

No. 4, pp. 693-712.

Teoh, S.H and T.J. Wong. 1993. Perceived auditor quality and the earnings response

coefficient. The Accounting Review. Vol. 68, No. 2, pp. 346-366.

Teoh, S.H., I. Welch, and T.J. Wong. 1998. Earnings management and the

underperformance of seasoned equity offerings. Journal of Financial Economics. Vol.

50, pp. 63-99.

Toms, S. and M. Wright. 2005. Divergence and convergence within Anglo-American

corporate governance systems: Evidence from the US and UK, 1950–2000. Business

History. Vol. 47, No. 2, pp. 267-295.

Titman, S. and B. Trueman. 1986. Information quality and the valuation of new

issues. Journal of Accounting and Economics. Vol. 8, pp. 159-172.

Treadway Commission. 1987. Report of the National Commission on Fraudulent

Financial Reporting. AICPA: New York.

Tsui, J.S.L., B. Jaggi, and F.A. Gul. 2001. CEO domination, growth opportunities and

their impact on audit fees. Journal of Accounting, Auditing and Finance. Vol. 16, No.

3, pp. 189–208

Turnbull, S. 1997. Corporate governance: Its scope, concerns and theories. Corporate

Governance. Vol. 5, No. 4, pp. 180-205.

The UK Corporate Governance Code. 2010. Financial Reporting Council: London.

Vafeas, N. 1999. Boar meeting frequency and firm performance. Journal of Financial

Vafeas, N. 2000. Board structure and the informativeness of earnings. Journal of

Accounting and Public Policy. Vol. 19, pp. 139-160.

Vafeas, N. 2005. Audit committees, boards, and the quality of reported earnings.

Contemporary Accounting Research. Vol. 22, No. 4, pp. 1093-1122.

Vance, S. 1983. Corporate leadership: Boards, directors, and strategy. McGraw-Hill:

New York.

Velury, U., J.T. Reisch and D. M. O‟Reilly. 2003. Institutional ownership and the

selection of industry specialist auditors. Review of Quantitative Finance and

Accounting. Vol. 21, pp. 35-48.

Walker, R.G. 2004. Gaps in guidelines on audit committees. ABACUS. Vol. 40, No. 2,

pp. 157-192.

Wallace, W.A. 1980. The economic role of the audit in free and regulated markets.

Touche Ross: New York

Wallman, S.M.H. 1996. The future of accounting, Part III: Reliability and auditor

independence. Accounting Horizons. Vol. 10, No. 4, pp. 76-97.

Warfield, T.D., J.J. Wild, and K.L. Wild. 1995. Managerial ownership, accounting

choices, and informativeness of earnings. Journal of Accounting and Economics. Vol.

20, pp.61-91.

Watts, R.L. and J.L. Zimmerman. 1978. Towards a positive theory of the

determination of accounting standards. The Accounting Review. Vol. 53, pp. 112-134.

Watts, R.L., and J.L. Zimmerman. 1983. Agency problems, auditing, and the theory

of the firm: Some Evidence. Journal of Law and Economics. Vol. XXVI, pp. 613-633

Watts, R.L., and J.L. Zimmerman. 1986. Positive accounting theory. Englewood

Cliffs, Prentice-Hall: NJ.

Watts, R.L., and J.L. Zimmerman. 1990. Positive accounting theory: A ten year

perspective. The Accounting Review. Vol. 65, No. 1, pp. 131-156.

Whisenant, S., S. Sankaraguruswamy and K. Raghunandan. 2003. Evidence on the

joint determination of audit and non-audit fees. Journal of Accounting Research. Vol.

41, No. 4, pp. 721-744.

Wines, G. 1994. Auditor independence, audit qualification and provision of non-audit

services: A note. Accounting and Finance. Vol. 34, No. 1, pp. 75-86.

Williams, R. 2009. Using heterogeneous choice models to compare logit and probit

coefficients across groups. Sociological Methods & Research. Vol. 37, No. 4, pp. 531-

Williams, R. 2010. Fitting heterogeneous choice models with oglm. The Stata

Journal. Vol. 10, No. 4, pp. 540-567.

Wilson, R. 1983. Auditing: Perspectives from multi-person decision theory. The

Accounting Review. Vo. LVIII, No. 2, pp.305-318.

Wilson, T.E. and R.A. Grimlund. 1990. An examination of the importance of an

auditor‟s reputation. Auditing: A Journal of Practice & Theory. Vol. 9, No. 2, pp. 43-

Willenborg, M. 1999. Empirical analysis of the economic demand for auditing in the

initial public offerings market. Journal of Accounting Research. Vol.37, No. 1, pp.

225–238.

Wolinsky, A. 1983. Prices as signals of product quality. Review of Economic Studies.

Vol. 50, No. 4, pp. 647-658.

Wolnizer, P.W. 1995. Are audit committee red herrings? ABACUS. Vol.31, No. 1, pp.

45-66.

Wooten, T.C. 2003. Research about audit quality. The CPA Journal. Available at

http://www.nysscpa.org/cpajournal/2003/0103/dept/d014803.htm

Wright, S., and A. Wright. 1997. The effect of industry specialization on hypothesis

generation and audit planning decision. Behavioural Research in Accounting. Vol. 9,

pp. 273-294.

Xie, B., W.N. Davidson, and P.J. DaDalt. 2003. Earnings management and corporate

governance: The Role of Board and the Audit Committee. Journal of Corporate

Finance. Vol. 9, No. 3, pp. 295-316

Yang, J.S., J. Krishnan. 2005. Audit committee and quarterly earnings management.

International Journal of Auditing. Vol. 9, pp. 201-219.

Yeoh, E., and C.A. Jubb. 2001. Governance and audit quality: Is there an association?

Working paper, The University of Melbourne.

Yermack, D. 1996. Higher market valuation of companies a small board of directors,

Journal of Financial Economics. Vol. 40, pp. 185-202.

Young, S. 1999. Systematic measurement error in estimation of accruals: An

evaluation of alternative modelling procedures. Journal of Business Finance &

Accounting. Vol. 26, No. 7/ 8, pp. 833-862.

Zahra, S.A., and J.A. Pearce. 1989. Boards of directors and corporate financial

performance: A review and integrative model. Journal of Management. Vol. 15, No.

2, pp. 291-334.

Zaman, M. M. Hudiab and R. Haniffa. 2011. Corporate governance quality, audit fees

and non-audit fees, Journal of Business Finance & Accounting, Vol. 38, No. 1 & 2,

pp. 165-197.

Appendix 1

The prohibitions of the specific non-audit services as outlined in ES 5 (revised) –

NAS provided to audited entities

INTERNAL AUDIT SERVICES

44 The audit firm shall not undertake an engagement to provide internal audit

services to an audited entity where it is reasonably foreseeable that:

(a) For the purposes of the audit of the financial statements, the auditor would

place significant reliance on the internal audit work performed by the

audit firm; or

(b) For the purposes of the internal audit services, the audit firm would

undertake part of the role of management.

INFORMATION TECHNOLOGY SERVICES

52 The audit firm shall not undertake an engagement to design, provide or

implement information technology systems for an audited entity where:

(a) the systems concerned would be important to any significant part of the

accounting system or to the production of the financial statements and the

auditor would place significant reliance upon them as part of the audit of

the financial statements; or

(b) For the purposes of the information technology services, the audit firm

would undertake part of the role of management.

VALUATION SERVICES

56 The audit firm shall not undertake an engagement to provide a valuation to:

(a) an audited entity that is a listed company or a significant affiliate of such

an entity, where the valuation would have a material effect on the listed

company‟s financial statements, either separately or in aggregate with

other valuations provided; or

(b) Any other audited entity, where the valuation would both involve a

significant degree of subjective judgment and have a material effect on the

financial statements either separately or in aggregate with other valuations

provided.

ACTUARIAL VALUATION SERVICES

63 The audit firm shall not undertake an engagement to provide actuarial

valuation services to:

(a) An audited entity that is a listed company or a significant affiliate of such

an entity, unless the firm is satisfied that the valuation has no material

effect on the listed company‟s financial statements, either separately or in

aggregate with other valuations provided; or

(b) any other audited entity, unless the firm is satisfied that either all

significant judgments, including the assumptions, are made by informed

management or the valuation has no material effect on the financial

statements, either separately or in aggregate with other valuations

provided.

TAX SERVICES

72 The audit firm shall not promote tax structures or products or undertake an

engagement to provide tax advice to an audited entity where the audit

engagement partner has, or ought to have, reasonable doubt as to the

appropriateness of the related accounting treatment involved, having regard to

the requirement for the financial statements to give a true and fair view in

accordance with the relevant financial reporting framework.

74 The audit firm shall not undertake an engagement to provide tax services

wholly or partly on a contingent fee basis where:

(a) the services are provided to an audited entity and the engagement fees are

material to the audit firm or the part of the firm by reference to which the

audit engagement partner‟s profit share is calculated; or

(b) The outcome of those tax services (and, therefore, the amount of the fee)

is dependent on:

(i) The application of tax law which is uncertain or has not been

established; and

(ii) A future or contemporary audit judgment relating to a material matter

in the financial statements of an audited entity.

76 The audit firm shall not undertake an engagement to provide tax services to an

audited entity where the engagement would involve the audit firm undertaking

a management role.

78 For an audited entity that is a listed company or a significant affiliate of such

an entity, the audit firm shall not undertake an engagement to prepare current

or deferred tax calculations for the purpose of preparing accounting entries

that are material to the relevant financial statements, save where the

circumstances contemplated in paragraph 131 apply.

82 The audit firm shall not undertake an engagement to provide tax services to an

audited entity where this would involve acting as an advocate for the audited

entity, before an appeals tribunal or court5 in the resolution of an issue:

(a) that is material to the financial statements; or

(b) where the outcome of the tax issue is dependent on a future or

contemporary audit judgment.

LITIGATION SUPPORT SERVICES

88 The audit firm shall not undertake an engagement to provide litigation support

services to:

an entity, where this would involve the estimation by the audit firm of the

likely outcome of a pending legal matter that could be material to the

amounts to be included or the disclosures to be made in the listed

company‟s financial statements, either separately or in aggregate with

other estimates and valuations provided; or

(b) any other audited entity, where this would involve the estimation by the

audit firm of the likely outcome of a pending legal matter that could be

material to the amounts to be included or the disclosures to be made in the

Consultation on audit firms providing non-audit services to listed

companies that they audit financial statements, either separately or in

aggregate with other estimates and valuations provided and there is a

significant degree of subjectivity involved.

LEGAL SERVICES

91 The audit firm shall not undertake an engagement to provide legal services to

an audited entity where this would involve acting as the solicitor formally

nominated to represent the audited entity in the resolution of a dispute or

litigation which is material to the amounts to be included or the disclosures to

be made in the financial statements.

RECRUITMENT AND REMUNERATION SERVICES

93 The audit firm shall not undertake an engagement to provide recruitment

services to an audited entity that would involve the firm taking responsibility

for the appointment of any director or employee of the audited entity.

95 For an audited entity that is a listed company, the audit firm shall not

undertake an engagement to provide recruitment services in relation to a key

management position of the audited entity, or a significant affiliate of such an

entity.

99 The audit firm shall not undertake an engagement to provide advice on the

quantum of the remuneration package or the measurement criteria on which

the quantum is calculated, for a director or key management position of an

audited entity.

CORPORATE FINANCE SERVICES

109 The audit firm shall not undertake an engagement to provide corporate finance

services in respect of an audited entity where:

(a) the engagement would involve the audit firm taking responsibility for

dealing in, underwriting or promoting shares; or

(b) the audit engagement partner has, or ought to have, reasonable doubt as to

the appropriateness of an accounting treatment that is related to the advice

provided, having regard to the requirement for the financial statements to

give a true and fair view in accordance with the relevant financial

reporting framework; or

(c) such corporate finance services are to be provided on a contingent fee

basis and:

(i) the engagement fees are material to the audit firm or the part of the

firm by reference to which the audit engagement partner‟s profit

share is calculated; or

(ii) the outcome of those corporate finance services (and, therefore, the

amount of the fee) is dependent on a future or contemporary audit

judgment relating to a material matter in the financial statements of

an audited entity; or

(d) the engagement would involve the audit firm undertaking a management

role in the audited entity.

TRANSACTION RELATED SERVICES

119 The audit firm shall not undertake an engagement to provide transaction

related services in respect of an audited entity where:

(a) the audit engagement partner has, or ought to have, reasonable doubt as to

the appropriateness of an accounting treatment that is related to the advice

provided, having regard to the requirement for the financial statements to

give a true and fair view in accordance with the relevant financial

reporting framework; or

(b) such transaction related services are to be provided on a contingent fee

basis and:

(i) the engagement fees are material to the audit firm or the part of the

firm by reference to which the audit engagement partner‟s profit

share is calculated; or

(ii) the outcome of those transaction related services (and, therefore,

the amount of the fee) is dependent on a future or contemporary

audit judgment relating to a material matter in the financial

statements of an audited entity; or

(c) the engagement would involve the audit firm undertaking a management

role in the audited entity.

ACCOUNTING SERVICES

127 The audit firm shall not undertake an engagement to provide accounting

services to:

an entity, save where the circumstances contemplated in paragraph 131

apply; or

(b) any other audited entity, where those accounting services would involve

the audit firm undertaking part of the role of management.

thesis rohaida plus corrections

nonaudit fees

audit committee

audit judgement

higher audit fees

perceived audit quality

proxies of audit quality

nonaudit services

durham theses

Documents

corrections and analysis gravity corrections

gamma camera (corrections). image sans corrections

corrections · inspector of corrections trevor longmuir...

thesis final with corrections - cardiff...

probe softwareprobesoftware.com/download/probesoftware...

phd thesis (with examiner corrections)...acknowledgement...

josefine's corrections - complete thesis...

justine jowell masters thesis final with corrections ·...

dc department of corrections media kit of columbia...

radiative corrections as the origin of … · radiative...

department of corrections overview · 2019-04-10 ·...

epubs.surrey.ac.ukepubs.surrey.ac.uk/841456/1/thesis final...

comments and corrections - daniel liberzon's...

dissertation and thesis manual€¦ · dissertation or...

corrections today eco freindly corrections-green...

phd thesis of gavin sutton with viva corrections july 06 _2

february 13, 2015 corrections of corrections of

dissertation docteur de l’universitÉ du...

center on sentencing and corrections the … · center on...

research design (part 2) siti rohaida bt mohamed zainal, phd...